session-hub
Skill1 starsChanged 21 days agoMITView on GitHub
Coordinate parallel agent sessions via a shared append-only HTML hub. Use on $session-hub, 'run parallel sessions on this', joining a hub, or proactively when edits or commits this session did not make appear in the checkout.
We split this skill into the rules it gives an agent
4 rules. About 0 are common, 2 are distinctive, and 2 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈2
Distinctive · estimate
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 14
Use when the effort splits into scopes different sessions can own without touching the same files: separate directories, repos, or lanes …
- Very specific: line 19
Never open a second hub for an effort.
- Very specific: line 42
Join: Read the whole hub first. Then add one roster row before the ROSTER anchor: session role name, start time, scope claimed, branch or …
- Distinctive: line 63
Cross-session messages: If the current session exposes a messaging tool that reaches another participant, use it for urgent interrupts only.
≈ “Common” is an estimate: at least 1 in 5 of the 308,232 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Shared ledger for parallel sessions
One effort, several sessions (Codex, Claude Code, or both), one hub file every session reads and appends to. The hub is the durable record: who is working, what scope each session owns, what happened when. The user leaves it open in a browser and watches all lanes at once.
When to use, when not
- Use when the effort splits into scopes different sessions can own without touching the same files: separate directories, repos, or lanes (build vs docs vs verification tooling).
- Skip when all sessions would edit the same few files. Parallelism there loses to merge conflicts; run serial instead and say so.
- Skip for single-session tasks. The hub earns its overhead only with 2+ concurrent sessions.
- Never open a second hub for an effort. Before creating one, check the established effort
location (Desktop or authorized workspace) for an existing
HUB-*.htmland join it.
Hub file
- Location: the user's Desktop when the sandbox allows access and the user authorized it;
otherwise an authorized workspace location. Share its absolute path with every participant.
Name it
HUB-<effort-slug>.html, one per effort. - Self-contained HTML with
<meta http-equiv="refresh" content="15">so an open browser tab updates itself. No external assets, no scripts required. - Created once from template.html in this skill's directory (fill the
EFFORT_TITLEplaceholder). Never rewrite the file wholesale after creation. - Two permanent anchors:
<!-- HUB:ROSTER:END -->and<!-- HUB:LOG:END -->. Every write is an insert immediately before one of them. Append-only: never edit or delete another session's entries, including typos; correct with a new entry.
The hub is a cooperative ownership ledger, not an OS lock or atomic transaction. Use exposed messaging and nonoverlapping file ownership; verify inserts survived after writing. If concurrent writes keep colliding, designate one writer or serialize updates. Do not claim the HTML enforces exclusivity.
Protocol
- Join. Read the whole hub first. Then add one roster row before the ROSTER anchor:
session role name, start time, scope claimed, branch or worktree. Pick a short role name
(
turret-build,docs-pass,verify-rig), not a model name; it is the session's identity in every entry. - Scope claims are exclusive. Claim explicit paths or globs. Touching a file inside
another session's claim requires an
askentry in the log and that session'sokentry (or the user overriding in chat). This exclusivity is the point of the hub; without it parallel sessions corrupt each other's work. - Log entries go before the LOG anchor: local time, role name, one of
status/finding/ask/ok/done, then a terse line with paths or commit SHAs as evidence. Get the time from the shell (date +%H:%Min bash,Get-Date -Format HH:mmin PowerShell); never guess it. - Cadence. Re-read the hub at natural boundaries: before starting a new task, after each commit, before claiming new scope, before answering an ask. No polling loops unless the user asks for one.
- Finish. Post a
doneentry with verification evidence (command output, screenshot path, final SHA), then update your own roster row's status cell to done. Your roster row is the one exception to "never edit existing entries", and only its status cell. - Write races. Anchored inserts from two sessions can collide. If a patch no longer matches, re-read the file and retry once. If entries interleave out of time order, leave them; order within a minute does not matter.
- Cross-session messages. If the current session exposes a messaging tool that reaches another participant, use it for urgent interrupts only; the hub stays the record. Anything decided over a direct message gets a log entry too.
Joining a hub or claiming scope does not authorize commits, pushes, PRs, merges, or deploys.
Those follow the user's explicit requests and AGENTS.md.
Entry examples
<tr><td>14:02</td><td>turret-build</td><td>status</td><td>tracking clamp landed, commit 3f1c2aa, starting speed tuning</td></tr>
<tr><td>14:05</td><td>docs-pass</td><td>ask</td><td>turret-build: may I edit src/main.ts header comment? inside your claim</td></tr>
<tr><td>14:06</td><td>turret-build</td><td>ok</td><td>docs-pass: yes, header comment only</td></tr>
<tr><td>14:31</td><td>verify-rig</td><td>done</td><td>headed run clean, screenshot .tmp/verify/final.png, all checklist items pass</td></tr>Other versions5 projects use a version of this.
Word for word · 4
- ryanportfolio/Corewise.Academy★ 1100% same
- ryanportfolio/savetokens★ 1100% same
- ryanportfolio/STK★ 1100% same
- ryanportfolio/HarnessFirmware.com★ 0100% same
Edited · 0
None found.
Wrote their own version · 1
- Gateton/pi-session-hub★ 39same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
Used alongsideProjects with it also use compact-review.
Other files people use alongside this one.
Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
ryanportfolioSkillcompact-review
Use on /compact-review or right before /compact: review the session and return copy-ready custom instructions for /compact that keep what matters. Does not run /compact.
Also in 4 projects that use this
In this repository85 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- addskillSkill
- adopt-repoSkill
- advocateSkill
- apply-firmwareSkill
- arenaSkill
- astra-reviewSkill
- babysit-ciSkill
- brainstormingSkill
- broSkill
- cache-loopSkill
- cavemanSkill
- claude-reviewSkill
- codex-reviewSkill
- dareSkill
- deep-planSkill
- design-prototypesSkill
- enhance-promptSkill
- external-reviewSkill
- fable-modeSkill
- forge-repo-ui-skillSkill
- handoff-auditSkill
- impartial-reviewSkill
- init-projectSkill
- labSkill
- long-horizonSkill
- optimize-contextSkill
- opus-fullreviewSkill
- perf-loopSkill
- recallSkill
- redesign-conceptsSkill
- refineSkill
- serversSkill
- showpieceSkill
- sync-starterSkill
- whySkill
- wow-loopSkill
- wrapupSkill
- writing-plansSkill
- writingSkill
- addskillSkill
- adopt-repoSkill
- advocateSkill
- apply-firmwareSkill
- arenaSkill
- astra-fullreviewSkill
- astra-reviewSkill
- babysit-ciSkill
- brainstormingSkill
- broSkill
- cache-loopSkill
- cavemanSkill
- claude-reviewSkill
- codex-fullreviewSkill
- codex-image-genSkill
- codex-reviewSkill
- coordinateSkill
- dareSkill
- deep-planSkill
- The first 60 shown.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
Reviews can't be read right now.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

