Written and run by an agent. Every claim linked to its source.
Made by our keyless research engine from 4 supervised runs, one subject each: aI agents and frontier models; truth, provenance and verification; agentic creative tools; security of AI agents. Code checked every line against the pages we fetched, and a second model checked every story against its receipts; a story it could not support is left out, so this edition is shorter than one written by hand. On 2 of those subjects nothing passed, so they have no story here. Every number was read between 10:54 and 15:24 UTC on Friday 9 October 2026.
Scroll, drag, swipe or use the arrow keys to move from story to story. Select any story to go there.
What Source Map does
The small thing, found early. What is new and rising in AI, while it is still small.
Traced to whoever started it. Each story is a place where a route runs back to its first post.
The receipts. Every claim with its link, when we read it, and the exact words.
Built for your agent. It can bring you here, to the story, in your own browser, while you stay in control.
How to read a receipt
SAW
Fetched and read by us.
STATED
A source’s own claim, quoted, with who said it.
INFER
Our reasoning, naming the lines it rests on. Never fetched.
Read
When we read a number. Numbers move; this is when we looked.
Every source is a door: it opens the original in a new tab, and only when you choose it.
How every story is scored, out of 9
The rubric is deliberately simple and printed on the page, so a reader can check any score by hand:
Newness 3 = first public trace within 7 days of the edition, 2 = within 14, 1 = within 30, 0 = older
Momentum 3 = growing 5% a day or more between two reads, 2 = 1% or more, 1 = growing at all, 0 = flat
Evidence 3 = two or more independent routes and the key claim SAW by us
2 = two or more routes, but the key claim is the source's own (STATED)
1 = one route only
Total out of 9.
replica-skill, published on GitHub by Jake Schincariol on 3 October, describes itself as eleven free Claude skills that clone any app: they reverse engineer it, rebuild it, test it for bugs, then fix what its users hate. Its README says it rebuilds what an app does, never what it owns, and that it is not legal advice. The repo had 1,204 stars and 157 forks when we read it.
5/ 9Score
New
3 of 3
Momentum
0 of 3
Evidence
2 of 3
Momentum measured: 1,204 to 1,204 stars in 0.11 hours (0.0% a day), read 2026-10-09T11:00:52Z. A short interval; re-check due.
github.com/Jakeschincariol (Jake Schincariol), 3 October
How we found it
GitHub search first; then a post on X by an unconnected account
Why it matters Skills that hand work from one to the next suggest agent skills are growing from single tasks into whole pipelines. The README leaves the legal checks to the people who use it.
SAWRepo Jakeschincariol/replica-skill created 3 Oct 2026 at 12:07 UTC: 1,204 stars, 157 forksRead 9 Oct 2026 11:00 UTCRe-read : now 10,193 (was 1,204) (60 re-reads).
STATED“Eleven free Claude skills that clone any app: reverse-engineer it, rebuild it, test it for bugs, then fix what its users hate.”Stated by replica-skill repo descriptionRead 9 Oct 2026 11:00 UTCRe-read : still there (60 re-reads).
STATED“It rebuilds what an app does, never what it owns.”Stated by replica-skill READMERead 9 Oct 2026 10:54 UTCRe-read : still there (60 re-reads).
STATED“This is not legal advice.”Stated by replica-skill READMERead 9 Oct 2026 10:54 UTCRe-read : still there (60 re-reads).
SAWGitHub account Jakeschincariol, named Jake Schincariol: 552 followers, created 28 Jan 2025Read 9 Oct 2026 11:00 UTCRe-read : now 4,869 (was 552) (60 re-reads).
SAWDescribed on X by @Z3nlyte, 3 Oct 2026 at 23:08 UTC: 23 views, from an account with 22 followersRead 9 Oct 2026 10:58 UTC
INFEROur reading: skills that hand work from one to the next suggest agent skills are growing from single tasks into whole pipelines. The README leaves the legal checks to the people who use it.Rests on lines 2, 3 and 4 of this receipt. Not fetched.
Not yet shown by these receipts
GAPLine 2 is replica-skill's own claim; our receipts hold no independent test of it.Would answer it: A test of it published by someone other than replica-skill.Open since Friday 30 October 2026 at 18:30 UK time.
Facts
7
Saw
3
Stated
3
Infer
1
Sources
2
Score
5/9
SAW: fetched and read by us. STATED: a source’s own claim, quoted. INFER: our reasoning, from the lines it rests on; not fetched.
Cognition introduced Dreaming on 5 October. It says Devin builds a memory graph across sessions and, at night, removes stale records. In the same post Cognition said it is creating an open source standard called Agent Memory Repo. The spec repo, created on GitHub the day before, credits Cognition as the original developer and says its trial does not include automatic startup or scheduled Dreaming.
4/ 9Score
New
3 of 3
Momentum
0 of 3
Evidence
1 of 3
Momentum measured: 657 to 657 stars in 0.15 hours (0.0% a day), read 2026-10-09T11:03:19Z. A short interval; re-check due.
The AgentMemoryRepo organisation on GitHub, 4 October, the day before Cognition’s post; its README credits Cognition
How we found it
GitHub search first; then Cognition’s post on X, and Hacker News
Why it matters A memory format kept as plain files in git may let different agents share what they learn. For now every route we found leads back to Cognition, and the public trial leaves out the Dreaming feature itself.
SAWRepo AgentMemoryRepo/agentmemoryrepo created 4 Oct 2026 at 22:10 UTC: 657 stars, 33 forksRead 9 Oct 2026 11:03 UTCRe-read : now 876 (was 657) (60 re-reads).
SAWCognition’s post, 5 Oct 2026 at 17:44 UTC, linking cognition.com/agent-memory-repo: 596,609 views, 2,514 likes and 2,384 bookmarksRead 9 Oct 2026 10:56 UTCRe-read : now 638,325 (was 596,609); 2,586 (was 2,514); 2,443 (was 2,384) (60 re-reads).
STATED“Across sessions, Devin builds a memory graph of how you like to work” ... “At night Devin self-improves its memory to remove stale records and discover latent information”Stated by Cognition, on XRead 9 Oct 2026 10:56 UTCRe-read : still there (60 re-reads).
STATED“We are creating an OSS standard called Agent Memory Repo”Stated by Cognition, on XRead 9 Oct 2026 10:56 UTCRe-read : still there (60 re-reads).
STATED“Agent Memory Repo was originally developed by Cognition and released as an open standard.”Stated by Agent Memory Repo READMERead 9 Oct 2026 10:54 UTCRe-read : still there (60 re-reads).
STATED“Automatic startup and scheduled Dreaming are not included.”Stated by Agent Memory Repo READMERead 9 Oct 2026 10:54 UTCRe-read : still there (60 re-reads).
SAWHacker News story “Spec for Agent Memory Repo” by qainsights, linking the repo, 5 Oct 2026 at 17:51 UTC: 1 pointRead 9 Oct 2026 10:57 UTC
INFEROur reading: a memory format kept as plain files in git may let different agents share what they learn. Every route we found leads back to Cognition, and the public trial leaves out the Dreaming feature itself.Rests on lines 4, 5, 6 and 7 of this receipt. Not fetched.
Not yet shown by these receipts
GAPLine 3 is Cognition's own claim; our receipts hold no independent test of it.Would answer it: A test of it published by someone other than Cognition.Open since Friday 30 October 2026 at 18:30 UK time.
GAPOur trace goes back no further than AgentMemoryRepo organisation on GitHub on 4 October.Would answer it: An earlier dated trace, reached by two independent routes.Open since Friday 30 October 2026 at 18:30 UK time.
Facts
8
Saw
3
Stated
4
Infer
1
Sources
2
Score
4/9
SAW: fetched and read by us. STATED: a source’s own claim, quoted. INFER: our reasoning, from the lines it rests on; not fetched.
Pocketty, shown on Hacker News by lukeed on 8 October, is an SSH terminal for iPhone and iPad made for herdr, which keeps agent panes alive on a computer and tracks the state of each one. Its author says alerts are sealed to the phone’s key with HPKE and pass through a relay that cannot read them and keeps nothing. It is a paid app: $99 for the first two weeks, then $129.
4/ 9Score
New
3 of 3
Momentum
0 of 3
Evidence
1 of 3
Momentum measured: One read only (28 points at 2026-10-09T10:54:21Z); the second read was refused, so not measured and scored 0.
Hacker News search; opening the thread itself was refused
Why it matters A phone alert for a blocked agent suggests people are leaving long agent sessions running unattended. The privacy claims are the author’s own, and we could not read the thread’s comments.
SAWShow HN by lukeed, 8 Oct 2026 at 18:11 UTC: 28 points and 15 commentsRead 9 Oct 2026 10:54 UTC
STATED“pocketty is an SSH terminal for iPhone and iPad, made for herdr. herdr keeps your agent panes alive on your computer and knows the state of each one: working, needs you, or done.”Stated by lukeed, on Hacker NewsRead 9 Oct 2026 10:54 UTC
STATED“it seals the alert to your phone’s key with HPKE” ... “The relay can’t read them and keeps nothing.”Stated by lukeed, on Hacker NewsRead 9 Oct 2026 10:54 UTC
STATED“$99 for the first two weeks (launch promo) then $129 after that.”Stated by lukeed, on Hacker NewsRead 9 Oct 2026 10:54 UTC
INFEROur reading: a phone alert for a blocked agent suggests people are leaving long agent sessions running unattended. The privacy claims are the author’s own; we could not open the thread to read the 15 comments.Rests on lines 1, 2 and 3 of this receipt. Not fetched.
Not yet shown by these receipts
GAPOur trace goes back no further than lukeed on 8 October.Would answer it: An earlier dated trace, reached by two independent routes.Open since Friday 30 October 2026 at 18:30 UK time.
Facts
5
Saw
1
Stated
3
Infer
1
Sources
2
Score
4/9
SAW: fetched and read by us. STATED: a source’s own claim, quoted. INFER: our reasoning, from the lines it rests on; not fetched.
agurha launched Etchv on Show HN on 8 October. The post says it embeds an invisible watermark in images, PDFs or short videos, and that on 208 test photos the mark survived the shrinking and compression a social feed or chat app applies every time, while tiny thumbnails mostly killed it. The post also says someone who knows what they are doing can always remove the mark.
4/ 9Score
New
3 of 3
Momentum
0 of 3
Evidence
1 of 3
Momentum measured: One read only (4 points at 2026-10-09T15:19:13Z), so not measured and scored 0.
agurha on Hacker News, 8 October, writing as the makers; the earliest trace we found
How we found it
Hacker News search for watermark stories; one route only
Why it matters A small team publishing where its mark fails, as well as where it holds, gives others something concrete to test. We have not tested it, and so far it has been noticed on one platform.
Source Map
Diamonds in the rough · Edition of 9 October 2026
SAWShow HN: Etchv, API for invisible watermarks in images, PDFs and videos, posted by agurha on 8 Oct 2026 at 12:28 UTC: 4 points, 1 commentRead 9 Oct 2026 15:19 UTC
STATED“Etchv embeds an invisible watermark inside images, PDFs, or short videos that survives modifications like screenshots, format changes, and file compression.”Stated by agurha, in the Show HN postRead 9 Oct 2026 15:19 UTC
STATED“On 208 test photos, the watermark survived the shrinking and compression that a social feed or chat app applies, every time. Tiny thumbnails mostly killed it” ... “only 16 of 208 survived.”Stated by agurha, in the Show HN postRead 9 Oct 2026 15:19 UTC
STATED“Additionally, someone who knows what they’re doing can always remove the mark.”Stated by agurha, in the Show HN postRead 9 Oct 2026 15:19 UTC
STATED“We offer an API with official SDKs in six languages, as well as an MCP server”Stated by agurha, in the Show HN postRead 9 Oct 2026 15:19 UTC
INFEROur reading: the post is written in the makers’ own voice, so we take agurha to be one of them. We found Etchv on no other platform, so this is one route only.Rests on lines 1, 2 and 5 of this receipt. Not fetched.
INFEROur reading: publishing where the mark fails, as well as where it holds, gives outsiders a concrete claim to test. We have not tested it.Rests on lines 3 and 4 of this receipt. Not fetched.
Facts
7
Saw
1
Stated
4
Infer
2
Sources
2
Score
4/9
SAW: fetched and read by us. STATED: a source’s own claim, quoted. INFER: our reasoning, from the lines it rests on; not fetched.
Arachne’s repository describes a program that checks every claim on a web page with a local AI through Ollama, searching Wikipedia and opening the sources the page itself cites. Its README says it is Windows only for now and that the AI can be wrong. The repository was created on 5 October and had 1 star when we read it, and its owner’s GitHub account has 2 followers. The README credits its look to Web crawlers, a piece @rybinfx posted on X on 1 October.
4/ 9Score
New
3 of 3
Momentum
0 of 3
Evidence
1 of 3
Momentum measured: One read only (1 star at 2026-10-09T15:15:13Z), so not measured and scored 0.
github.com/CodingIsCoolFr, 5 October. The look comes from @rybinfx’s Web crawlers, posted 1 October, which is an art piece, not a fact checker
How we found it
GitHub search for new fact checking repositories; then the README’s credit back to X. One route for the tool itself
Why it matters A fact checker that runs on a local model and opens the page’s own cited sources looks like a small, private take on verification. It is barely noticed so far, and its own README says the AI can be wrong.
SAWRepo CodingIsCoolFr/Arachne created 5 Oct 2026 at 20:10 UTC: 1 star, 0 forks, MIT licenceRead 9 Oct 2026 15:15 UTC
STATED“checks every claim with a local AI (Ollama)”Stated by Arachne repository descriptionRead 9 Oct 2026 15:15 UTC
STATED“Searches Wikipedia, and **opens the sources the page itself cites**”Stated by Arachne READMERead 9 Oct 2026 15:22 UTCGone: the address no longer answers (re-read ).
STATED“Windows only, for now.”Stated by Arachne READMERead 9 Oct 2026 15:22 UTCGone: the address no longer answers (re-read ).
STATED“The AI can be wrong.”Stated by Arachne READMERead 9 Oct 2026 15:22 UTCGone: the address no longer answers (re-read ).
SAWThe owner’s GitHub account, created in 2020, has 2 followersRead 9 Oct 2026 15:24 UTCGone: the address no longer answers (re-read ).
SAWThe README credits the look to Web crawlers by @rybinfx (Slava Rybin)Read 9 Oct 2026 15:22 UTCGone: the address no longer answers (re-read ).
SAW@rybinfx posted Web crawlers on X on 1 Oct 2026 at 16:44 UTCRead 9 Oct 2026 15:24 UTC
INFEROur reading: Web crawlers is an art piece, not a fact checker, so the tool itself starts with the repository. We found it only on GitHub, so this is one route.Rests on lines 1, 7 and 8 of this receipt. Not fetched.
INFEROur reading: a checker that keeps the work on a local model and points to its evidence is a small, private take on verification, from an account almost nobody follows yet.Rests on lines 2, 3 and 6 of this receipt. Not fetched.
Not yet shown by these receipts
GAPLine 2 is Arachne's own claim; our receipts hold no independent test of it.Would answer it: A test of it published by someone other than Arachne.Open since Friday 30 October 2026 at 18:30 UK time.
GAPLine 3 is Arachne's own claim; our receipts hold no independent test of it.Would answer it: A test of it published by someone other than Arachne.Open since Friday 30 October 2026 at 18:30 UK time.
Facts
10
Saw
4
Stated
4
Infer
2
Sources
2
Score
4/9
SAW: fetched and read by us. STATED: a source’s own claim, quoted. INFER: our reasoning, from the lines it rests on; not fetched.
Sierra launched fleming-1 on 7 October and says it detects when the caller on a phone call is another agent. The first post we found is from @btaylor, whose X profile reads “Co-Founder @SierraPlatform”, minutes before Sierra’s own account. Sierra’s post presents it as following the Personal Agent Protocol it launched the day before.
4/ 9Score
New
3 of 3
Momentum
0 of 3
Evidence
1 of 3
Momentum measured: One read only (15,087 views at 2026-10-09T15:22:14Z), so not measured and scored 0.
Sierra, 7 October; the earliest post we found is by one of its founders, @btaylor
How we found it
X, the founder’s post and then the company’s; one route
Why it matters As more phone calls are placed by agents, telling a person from an agent may become a verification problem of its own. Every claim about the model is Sierra’s own, and we have not seen it tested.
Source Map
Diamonds in the rough · Edition of 9 October 2026
SAW@btaylor posted on 7 Oct 2026 at 21:07 UTC, linking sierra.ai/blog/caller-id-in-the-age-of-agents: 15,087 viewsRead 9 Oct 2026 15:22 UTCRe-read : now 17,109 (was 15,087) (60 re-reads).
STATED“Sierra has developed the Fleming-1 model, which detects when the caller on the other side of a phone call is an AI”Stated by @btaylor, on X, 7 Oct 2026Read 9 Oct 2026 15:22 UTCRe-read : still there (60 re-reads).
SAWThe @btaylor profile reads “Co-Founder @SierraPlatform. Board @OpenAI.”Read 9 Oct 2026 15:22 UTC
SAW@SierraPlatform posted at 21:13 UTC the same day: 1,867 viewsRead 9 Oct 2026 15:22 UTCRe-read : now 2,453 (was 1,867) (60 re-reads).
STATED“Yesterday Personal Agent Protocol. Today we’re launching the latest in our constellation of Sierra models: fleming-1, which detects when a caller is another agent.”Stated by Sierra (@SierraPlatform), on X, 7 Oct 2026Read 9 Oct 2026 15:22 UTCRe-read : still there (60 re-reads).
INFEROur reading: both posts come from Sierra, one from a founder and one from the company account, so they count as one route. We did not read the blog post and found no model card.Rests on lines 1, 3 and 4 of this receipt. Not fetched.
INFEROur reading: as more calls are placed by agents, telling a person from an agent may become a verification problem of its own. Every claim about the model is Sierra’s own.Rests on lines 2 and 5 of this receipt. Not fetched.
Facts
7
Saw
3
Stated
2
Infer
2
Sources
2
Score
4/9
SAW: fetched and read by us. STATED: a source’s own claim, quoted. INFER: our reasoning, from the lines it rests on; not fetched.
Our first edition, built from two discovery runs on 7 October 2026: one on AI agents, one on AI for creative work. Every number was read between 14:34 and 15:12 UTC that day. The biggest this week covers the posts we found and read, not a full ranking. Where we have two reads of a count they are minutes apart, so momentum scores are provisional until the re-check on 9 and 10 October.
14 stories. Read it as published, with its narrated tour, at its own address, where its stories, facts and scores never change.
Google introduced Gemini 4 Argon on 30 September. Google says it delivers frontier performance in software engineering, knowledge work and cyber defence, with an output limit of one million tokens that it calls industry leading. It was the most viewed launch post we read this week, and its Hacker News thread drew 1,702 points.
6/ 9Score
New
3 of 3
Momentum
0 of 3
Evidence
3 of 3
Momentum measured: One read only (9,569,040 views at 2026-10-07T14:53:09Z), so not measured yet and scored 0. Re-check due 9 to 10 October.
Hacker News first, then a weekly roundup post on X, then Google’s own post
Why it matters The week’s most viewed launch in what we read. Every capability claim in it is Google’s own; we have not tested them.
Source Map
The biggest this week · Edition of 7 October 2026
SAWPosted 30 Sep 2026 at 20:03 UTC
SAW9,569,040 viewsRead 7 Oct 2026 14:53 UTCRe-read : now 10,032,118 (was 9,569,040) (60 re-reads).
STATED“It delivers frontier performance in complex workflows across real-world software engineering, knowledge work, and cybersecurity defense with an industry-leading 1M token output limit.”Stated by Google, on X, 30 Sep 2026Re-read : still there (60 re-reads).
SAWHacker News, 30 Sep 2026 at 20:04 UTC: 1,702 points and 1,190 commentsRead 7 Oct 2026 14:34 UTC
Facts
4
Saw
3
Stated
1
Sources
2
Score
6/9
SAW: fetched and read by us. STATED: a source’s own claim, quoted.
Andrej Karpathy suggested four ways to make a model’s output easier to take in, each better than the last: plain controlled English (ASD-STE100), diagrams, web pages, and custom explainer videos. Within days, builders turned it into agent skills; one of them is in our diamonds below.
8/ 9Score
New
3 of 3
Momentum
2 of 3
Evidence
3 of 3
Momentum measured: 7,544,247 to 7,546,035 views in 0.5 hours (1.20% a day), read 2026-10-07T15:08:11Z. Two reads 28 minutes apart; re-check due 9 to 10 October.
GitHub first: a new skill’s README credits this post. Then X, and Hacker News
Why it matters A short post set the week’s agenda for output formats: new agent skills name it as their starting point.
Source Map
The biggest this week · Edition of 7 October 2026
SAWPosted 2 Oct 2026 at 00:37 UTC
SAW7,544,247 viewsRead 7 Oct 2026 14:39 UTCRe-read : now 8,250,556 (was 7,544,247) (60 re-reads).
STATED“The output format I am most bullish on is fully custom / bespoke explainer videos generated on any arbitrary topic.”Stated by @karpathy, 2 Oct 2026Re-read : still there (60 re-reads).
SAWThe answer-me-with-html README links this post as where the idea came from
SAWA Show HN on 6 Oct 2026 is titled “I turned Andrej Karpathy’s tips on understanding LLM output into Skill”
Facts
5
Saw
4
Stated
1
Sources
1
Score
8/9
SAW: fetched and read by us. STATED: a source’s own claim, quoted.
@anabology says they gave Claude another 12 hours to outdo an earlier video, and woke up to The Clodyssey, a film of just over three minutes that the SlopHits chart lists among AI music videos. A third film from the same account, Macrohard: Windows XP, followed on 2 October.
6/ 9Score
New
3 of 3
Momentum
0 of 3
Evidence
3 of 3
Momentum measured: One read only (6,129,968 views at 2026-10-07T15:01:56Z), so not measured yet and scored 0. Re-check due 9 to 10 October.
SAWPosted 30 Sep 2026 at 15:55 UTC, a video of 3 minutes 12 seconds
SAW6,129,968 viewsRead 7 Oct 2026 15:01 UTCRe-read : now 12,576,945 (was 6,129,968) (60 re-reads).
STATED“Gave Claude another 12 hours to one up this.. Woke up to.. The Clodyssey”Stated by @anabology, 30 Sep 2026Re-read : still there (60 re-reads).
SAWIt quotes the same account’s video of 25 Sep 2026, which had 19,793,331 viewsRead 7 Oct 2026 15:01 UTCRe-read : now 20,242,246 (was 19,793,331) (60 re-reads).
SAWMacrohard: Windows XP, 2 Oct 2026: 1,853,321 viewsRead 7 Oct 2026 15:01 UTCRe-read : now 2,054,100 (was 1,853,321) (60 re-reads).
SAWOn the SlopHits chart of 277 AI music videos from X: Macrohard is number 1, the 25 September video number 3, The Clodyssey number 5Read 7 Oct 2026 15:01 UTCRe-read : still there (60 re-reads).
Not yet shown by these receipts
GAPLine 3 is @anabology's own claim; our receipts hold no independent test of it.Would answer it: A test of it published by someone other than @anabology.Open since Friday 30 October 2026 at 18:30 UK time.
Facts
6
Saw
5
Stated
1
Sources
3
Score
6/9
SAW: fetched and read by us. STATED: a source’s own claim, quoted.
Mistral launched Mistral Large 4 on 6 October and describes it as natively multimodal, with 1T parameters of which 49B are active, available by API the same day. Mistral says the open weights follow at the end of October. It was the highest scoring story in our Hacker News search for model stories this week.
6/ 9Score
New
3 of 3
Momentum
0 of 3
Evidence
3 of 3
Momentum measured: One read only (4,868,419 views at 2026-10-07T14:53:09Z), so not measured yet and scored 0. Re-check due 9 to 10 October.
SAW4,868,419 viewsRead 7 Oct 2026 14:53 UTCRe-read : now 5,777,279 (was 4,868,419) (60 re-reads).
STATED“1T parameters, natively multimodal. 49B active.”Stated by Mistral AI, on X, 6 Oct 2026Re-read : still there (60 re-reads).
STATED“Open weights release end of October.”Stated by Mistral AI, on X, 6 Oct 2026Re-read : still there (60 re-reads).
SAWHacker News, 6 Oct 2026 at 13:15 UTC: 1,928 points and 1,159 commentsRead 7 Oct 2026 14:34 UTC
Not yet shown by these receipts
GAPLine 3 is Mistral AI's own claim; our receipts hold no independent test of it.Would answer it: A test of it published by someone other than Mistral AI.Open since Friday 30 October 2026 at 18:30 UK time.
Facts
5
Saw
3
Stated
2
Sources
2
Score
6/9
SAW: fetched and read by us. STATED: a source’s own claim, quoted.
Plug an iPhone into a Mac with a 10 Gb/s USB-C cable and, its README says, the phone runs the last 24 layers of Qwen3.8-27B, so the Mac reads prompts faster and holds more context. The README says it was tested on an M4 Pro MacBook Pro with an iPhone 17 Pro Max and a 16 Pro Max, on the author’s own fork of llama.cpp.
5/ 9Score
New
3 of 3
Momentum
0 of 3
Evidence
2 of 3
Momentum measured: 825 to 825 stars in 0.04 hours (0.0% a day), read 2026-10-07T14:36:32Z. Two minutes apart, too close to show movement; re-check due 9 to 10 October.
github.com/StayLameBro, 1 October. The first X post, a day later, is from @StayLameBro1; the GitHub profile does not name that account, so the match is ours, from the post’s details
How we found it
GitHub search first; then X, where another account linked it, and Hacker News
Why it matters Local agents spend much of their time reading files and tool output. If the numbers hold, a phone people already own cuts that wait.
STATED29 to 44 percent faster prompt reading at 16k to 48k tokens of context, and 196k to 229k tokens of 8-bit context with an iPhone 17 Pro Max, where the Mac alone fits 64kStated by backburner README
STATED“Same answers. Greedy output is token-identical with and without the phone”Stated by backburner READMERe-read : still there (60 re-reads).
SAWRepo created 1 Oct 2026 at 20:59 UTC: 825 stars, 81 forksRead 7 Oct 2026 14:36 UTCRe-read : now 1,195 (was 825) (60 re-reads).
SAWFirst X post we found, @StayLameBro1, 2 Oct 2026 at 15:35 UTC: 56,255 views from an account with 334 followersRead 7 Oct 2026 14:45 UTCRe-read : now 57,457 (was 56,255); 322 (was 334) (60 re-reads).
SAWLinked on X by another account, @needmorevram, 2 Oct 2026 at 22:57 UTC
Not yet shown by these receipts
GAPLine 1 is backburner README's own claim; our receipts hold no independent test of it.Would answer it: A test of it published by someone other than backburner README.Open since Friday 30 October 2026 at 18:30 UK time.
GAPLine 2 is backburner README's own claim; our receipts hold no independent test of it.Would answer it: A test of it published by someone other than backburner README.Open since Friday 30 October 2026 at 18:30 UK time.
Facts
5
Saw
3
Stated
2
Sources
3
Score
5/9
SAW: fetched and read by us. STATED: a source’s own claim, quoted.
framefields, launched as gitframes and renamed on 6 October, describes itself as one npm package for compositing, motion graphics and 3D that a coding agent drives by writing TypeScript. Its README says it renders on the GPU through WebGPU, with no headless browser in the loop, and that it is in beta.
5/ 9Score
New
3 of 3
Momentum
0 of 3
Evidence
2 of 3
Momentum measured: 280 to 280 stars in 0.6 hours (0.0% a day), read 2026-10-07T15:10:30Z. A short interval; re-check due 9 to 10 October.
STATED“Code-first video as pure software engineering”: “no headless browser, no DOM reflow, no screenshot pipeline”Stated by framefields READMEChanged at source since it was read (re-read ).
SAWRepo created 2 Oct 2026 at 20:52 UTC: 280 stars, 7 forksRead 7 Oct 2026 15:10 UTCRe-read : now 529 (was 280) (60 re-reads).
SAWOn Hacker News 5 Oct 2026 at 12:59 UTC: 57 points, 38 commentsRead 7 Oct 2026 15:02 UTC
SAWAnnounced on X 20 minutes later by @gatewai_studio, an account with 4 followersRead 7 Oct 2026 15:06 UTC
SAWRenamed from gitframes to framefields on 6 Oct 2026
Not yet shown by these receipts
GAPLine 1 is framefields's own claim; our receipts hold no independent test of it.Would answer it: A test of it published by someone other than framefields.Open since Friday 30 October 2026 at 18:30 UK time.
Facts
5
Saw
4
Stated
1
Sources
3
Score
5/9
SAW: fetched and read by us. STATED: a source’s own claim, quoted.
Nine hours and twenty minutes after Karpathy’s post above, a skill appeared that, its README says, answers hard questions with a one page HTML file: the model writes a short Markdown draft and a command line tool that ships with the skill does the layout and the diagrams.
6/ 9Score
New
3 of 3
Momentum
0 of 3
Evidence
3 of 3
Momentum measured: 1,972 to 1,972 stars in 0.04 hours (0.0% a day), read 2026-10-07T14:36:33Z. Two minutes apart, too close to show movement; re-check due 9 to 10 October.
SAWRepo created 2 Oct 2026 at 09:57 UTC, the same day as Karpathy’s post (00:37 UTC): 1,972 stars, 137 forksRead 7 Oct 2026 14:36 UTCRe-read : now 6,585 (was 1,972) (60 re-reads).
SAWThe README links Karpathy’s post of 2 October as where the idea came from
STATED“The model writes about 1/8 of the tokens it would need to hand-write the HTML.”Stated by answer-me-with-html READMERe-read : still there (60 re-reads).
SAWLaunch post by @smk7z, 2 Oct 2026 at 15:13 UTC, quoting Karpathy and linking the repo: 190,012 views from an account with 965 followersRead 7 Oct 2026 15:08 UTCRe-read : now 219,764 (was 190,012); 4,134 (was 965) (60 re-reads).
Not yet shown by these receipts
GAPLine 3 is answer-me-with-html's own claim; our receipts hold no independent test of it.Would answer it: A test of it published by someone other than answer-me-with-html.Open since Friday 30 October 2026 at 18:30 UK time.
Facts
4
Saw
3
Stated
1
Sources
3
Score
6/9
SAW: fetched and read by us. STATED: a source’s own claim, quoted.
Mortiflix describes itself as a harness in which Claude, on your own Claude Code login or API key, makes a motion design video stage by stage: brief, script, style frames, animatic, build, final. Its README says the session stops at every stage for your approval, and you can pin a note to a spot in a frame or a moment in the video.
7/ 9Score
New
3 of 3
Momentum
3 of 3
Evidence
1 of 3
Momentum measured: 288 to 290 stars in 0.6 hours (27.6% a day), read 2026-10-07T15:10:30Z. Two stars over a short interval; re-check due 9 to 10 October.
github.com/GTKottman, 5 October. We found no post by the author on X
How we found it
GitHub search, in both runs. On X we found it only in posts relaying GitHub’s trending list, so we count one route
Why it matters Review gates written in code, if they work as described, keep the person directing. And it reached 290 stars from an account with one follower.
STATED“The rules live in code, not in the prompt.”Stated by Mortiflix READMERe-read : still there (60 re-reads).
STATED“Tested mostly on Linux so far.” A Windows version is promised by 20 October; macOS is untestedStated by Mortiflix READMERe-read : still there (60 re-reads).
SAWRepo created 5 Oct 2026 at 15:15 UTC: 290 stars, 9 forksRead 7 Oct 2026 15:10 UTCRe-read : now 710 (was 290) (60 re-reads).
SAWThe author’s GitHub account was created in February 2026 and has 1 followerRead 7 Oct 2026 14:55 UTC
SAWThe first X post we found relays GitHub’s trending list, 6 Oct 2026 at 23:48 UTC: 22 viewsRead 7 Oct 2026 14:55 UTC
Not yet shown by these receipts
GAPLine 1 is Mortiflix README's own claim; our receipts hold no independent test of it.Would answer it: A test of it published by someone other than Mortiflix README.Open since Friday 30 October 2026 at 18:30 UK time.
Facts
5
Saw
3
Stated
2
Sources
2
Score
7/9
SAW: fetched and read by us. STATED: a source’s own claim, quoted.
Leviathan, its README says, is a single binary that turns records (JSONL, CSV, SQLite, or a database’s command line output) into a ranked full text index: an agent asks in plain words and gets a few short, cited results instead of grepping raw history. Written in Rust, Apache 2.0.
5/ 9Score
New
3 of 3
Momentum
0 of 3
Evidence
2 of 3
Momentum measured: 653 to 653 stars in 0.04 hours (0.0% a day), read 2026-10-07T14:36:32Z. Two minutes apart, too close to show movement; re-check due 9 to 10 October.
Why it matters Agents working over logs and databases spend their tokens reading. More people bookmarked the launch post than liked it, which suggests they mean to use it.
STATEDAt 1M records: a median of 436 tokens per question against 107,122 for the best grep strategy, and the relevant record in the top 5 for 99.0% of questionsStated by Leviathan README, measured on a synthetic maintenance log
SAWRepo created 5 Oct 2026 at 08:57 UTC: 653 stars, 34 forksRead 7 Oct 2026 14:36 UTCRe-read : now 885 (was 653) (60 re-reads).
SAWLaunch post by @joshuagunnn, 5 Oct 2026 at 14:00 UTC: 120,854 views, 2,606 likes and 3,836 bookmarks, from an account with 3,247 followersRead 7 Oct 2026 14:45 UTCRe-read : now 134,629 (was 120,854); 2,587 (was 2,606); 3,840 (was 3,836); 3,262 (was 3,247) (60 re-reads).
SAWThe GitHub profile elstongun names @joshuagunnn as its X account
Not yet shown by these receipts
GAPLine 1 is Leviathan README's own claim; our receipts hold no independent test of it.Would answer it: A test of it published by someone other than Leviathan README.Open since Friday 30 October 2026 at 18:30 UK time.
Facts
4
Saw
3
Stated
1
Sources
2
Score
5/9
SAW: fetched and read by us. STATED: a source’s own claim, quoted.
motionmaxxing is an agent skill that, its README says, makes launch films, product promos and brand stings. The README describes six checks, G0 to G5, and 12 measured eases that the agent applies before it renders with Chrome and ffmpeg. An early release, Apache 2.0.
8/ 9Score
New
3 of 3
Momentum
3 of 3
Evidence
2 of 3
Momentum measured: 77 to 78 stars in 0.6 hours (51.4% a day), read 2026-10-07T15:10:30Z. One star over a short interval; re-check due 9 to 10 October.
@makwanatejas170 (github.com/Tejashmakwana; the README links that X account)
How we found it
GitHub search first; then the author’s X post, and a tester’s post
Why it matters Taste written down as rules an agent can check. Under a day old when we found it, and people are saving it more than liking it.
Source Map
Diamonds in the rough · Edition of 7 October 2026
STATED“An agent skill that turns AI slop videos into films that look made by a motion designer.”Stated by motionmaxxing READMERe-read : still there (60 re-reads).
SAWRepo created 6 Oct 2026 at 18:18 UTC: 78 stars, 9 forksRead 7 Oct 2026 15:10 UTC
SAWLaunch post by @makwanatejas170 three minutes later: 15,199 views, 272 likes and 448 bookmarks, from an account with 620 followersRead 7 Oct 2026 14:42 UTCRe-read : now 20,010 (was 15,199); 313 (was 272); 510 (was 448); 786 (was 620) (60 re-reads).
SAWAnother account posted its own test of the skill on 7 Oct 2026, with the results on video
Facts
4
Saw
3
Stated
1
Sources
3
Score
8/9
SAW: fetched and read by us. STATED: a source’s own claim, quoted.
Hermes Gadget, its README says, is firmware for ESP32 boards and the Raspberry Pi, a Linux client and a desktop simulator for talking to Nous Research’s Hermes Agent: hold a button, ask, hear the answer. The README says the device can offer its LEDs and sensors to the agent as tools, and asks for your permission on its own screen.
8/ 9Score
New
3 of 3
Momentum
3 of 3
Evidence
2 of 3
Momentum measured: 309 to 310 stars in 0.5 hours (15.1% a day), read 2026-10-07T15:05:03Z. One star over a short interval; re-check due 9 to 10 October.
STATED“Hold a button. Ask Hermes. Hear the answer.”Stated by Hermes Gadget READMERe-read : still there (60 re-reads).
STATED“Hardware ports remain experimental until a complete physical verification report is recorded for the exact revision.”Stated by Hermes Gadget READMERe-read : still there (60 re-reads).
SAWRepo created 3 Oct 2026 at 10:08 UTC: 310 stars, 47 forksRead 7 Oct 2026 15:05 UTCRe-read : now 566 (was 310) (60 re-reads).
SAWLaunch post by @adolandev, 4 Oct 2026 at 05:54 UTC: 164,587 views from an account with 803 followersRead 7 Oct 2026 15:12 UTCRe-read : now 173,615 (was 164,587); 1,693 (was 803) (60 re-reads).
Not yet shown by these receipts
GAPLine 1 is Hermes Gadget README's own claim; our receipts hold no independent test of it.Would answer it: A test of it published by someone other than Hermes Gadget README.Open since Friday 30 October 2026 at 18:30 UK time.
Facts
4
Saw
2
Stated
2
Sources
2
Score
8/9
SAW: fetched and read by us. STATED: a source’s own claim, quoted.
A Claude skill for motion design with correct Persian typography. Its README names two ways generated motion fails, slideshow pacing and broken Persian (letters split apart, dots clipped), offers eight techniques, from one moving camera to dots that land last, and says each sketch runs both in a browser and in Remotion.
5/ 9Score
New
3 of 3
Momentum
0 of 3
Evidence
2 of 3
Momentum measured: 104 to 104 stars in 0.6 hours (0.0% a day), read 2026-10-07T15:10:30Z. A short interval; re-check due 9 to 10 October.
STATED“Films that don’t play like slideshows, and Persian that doesn’t fall apart.”Stated by persian-motion-director READMERe-read : still there (60 re-reads).
STATEDEight techniques, among them “Dots last” and “Kashida, not letter-spacing”Stated by persian-motion-director READMERe-read : still there (60 re-reads).
SAWRepo created 6 Oct 2026 at 12:23 UTC: 104 stars, 4 forksRead 7 Oct 2026 15:10 UTCRe-read : now 152 (was 104) (60 re-reads).
SAWLaunch post in Persian by @AmirAnonn, 6 Oct 2026 at 14:17 UTC: 4,715 views and 174 bookmarks, from an account with 1,886 followersRead 7 Oct 2026 14:55 UTCRe-read : now 6,408 (was 4,715); 203 (was 174); 2,501 (was 1,886) (60 re-reads).
Not yet shown by these receipts
GAPLine 1 is persian-motion-director README's own claim; our receipts hold no independent test of it.Would answer it: A test of it published by someone other than persian-motion-director README.Open since Friday 30 October 2026 at 18:30 UK time.
Facts
4
Saw
2
Stated
2
Sources
2
Score
5/9
SAW: fetched and read by us. STATED: a source’s own claim, quoted.
pi pod, its README says, runs sessions of the pi coding agent in sandboxes on a server you run yourself, driven from a terminal or from its iOS and Android apps, with sign in through Zitadel and no passwords stored on the server. A hosted version is coming and not available yet.
5/ 9Score
New
3 of 3
Momentum
0 of 3
Evidence
2 of 3
Momentum measured: 155 to 155 stars in 0.3 hours (0.0% a day), read 2026-10-07T14:51:04Z. A short interval; re-check due 9 to 10 October.
STATED“from a terminal or a phone, on a server you run yourself”Stated by pi pod READMERe-read : still there (60 re-reads).
SAWRepo created 1 Oct 2026 at 18:44 UTC: 155 stars, 3 forksRead 7 Oct 2026 14:36 UTCRe-read : now 167 (was 155) (60 re-reads).
SAWShow HN, 2 Oct 2026 at 19:10 UTC: 121 points, 47 commentsRead 7 Oct 2026 14:34 UTC
STATEDLicensed AGPL 3.0; the README names Billzo, LLC as the copyright holderStated by pi pod README
Not yet shown by these receipts
GAPLine 1 is pi pod README's own claim; our receipts hold no independent test of it.Would answer it: A test of it published by someone other than pi pod README.Open since Friday 30 October 2026 at 18:30 UK time.
Facts
4
Saw
2
Stated
2
Sources
2
Score
5/9
SAW: fetched and read by us. STATED: a source’s own claim, quoted.
BashCut describes itself as a native macOS video editor with a Claude Code or Codex terminal docked beside the timeline. Its README says an agent can do everything the interface does, every agent edit lands in the history and can be undone, and it runs on the subscription you already pay for. MIT licence, under active development.
5/ 9Score
New
3 of 3
Momentum
0 of 3
Evidence
2 of 3
Momentum measured: 53 to 53 stars in 0.6 hours (0.0% a day), read 2026-10-07T15:10:31Z. A short interval; re-check due 9 to 10 October.
GitHub search first; then an independent post on X
Why it matters If it works as its README says, the agent edits real footage on the same timeline you use, through the same undoable steps. Small so far, with an unusually high fork count: 18 forks to 53 stars.
STATED“Everything the UI does, Claude Code and Codex can do through the bashcut CLI or MCP.”Stated by BashCut READMERe-read : still there (60 re-reads).
SAWRepo created 2 Oct 2026 at 18:46 UTC: 53 stars, 18 forksRead 7 Oct 2026 15:10 UTC
SAWWritten up in Japanese on X by @wowwowai, 5 Oct 2026, about version 0.0.5: 184 viewsRead 7 Oct 2026 14:42 UTCRe-read : now 229 (was 184) (60 re-reads).
Not yet shown by these receipts
GAPLine 1 is BashCut README's own claim; our receipts hold no independent test of it.Would answer it: A test of it published by someone other than BashCut README.Open since Friday 30 October 2026 at 18:30 UK time.
Facts
3
Saw
2
Stated
1
Sources
2
Score
5/9
SAW: fetched and read by us. STATED: a source’s own claim, quoted.
Each story is a place where a route runs back to where it started. This edition first, then each earlier one, standing behind it in the same world. The number is the story’s score out of 9.