agent-bridge
Skill1 starsChanged 9 days agoMITView on GitHub
Use when the user gives a task for hand-off to a reviewer (Claude), says "handoff", "report", "publish report" or "bridge", or when .handoff/TASK.md exists. Does the task locally, writes a structured report, and can publish a sanitized copy of the report to the agent-bridge log repo so the reviewer can read it.
- Commits and pushes
10 other projects use a version of this.
Adopted by
Most recently taken up by
- onblueroses/stratatheir own version, first seen Nov 16, 2026
- kununu/agent-bridgetheir own version, first seen Nov 12, 2026
- avas-app/agent-bridgetheir own version, first seen Nov 12, 2026
- Latth/agent-bridgetheir own version, first seen Nov 8, 2026
- cocodrino/bridge-harnesstheir own version, first seen Oct 25, 2026
- ilamanov/skillstheir own version, first seen Oct 13, 2026
Other versions10 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 10
- kununu/agent-bridge★ 18same name
- xyva-yuangui/agent-bridge★ 15same name
- EthanSK/agent-bridge★ 11same name
- avas-app/agent-bridge★ 4same name
- cocodrino/bridge-harness★ 4same name
- onblueroses/strata★ 3same name
- Latth/agent-bridge★ 2same name
- ilamanov/skills★ 1same name
- JaylanJerry/codex-agent-bridge★ 0same name
- Mihirokte/agent-skills★ 0same 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.
The file
We split this skill into the rules it gives an agent
7 rules. 0 are common, 3 are distinctive, and 4 are very specific to this project.
0
Common
Most of the 9 edited versions we read keep these.
3
Distinctive
This author's own choices. Worth a look.
4
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- could be automatic: a setting could enforce it
- Distinctive: line 14
Read the repo first (README, manifests, tests, git log -5).
- Distinctive: line 16
Do not touch files outside the task.
- Very specific: line 36
Run git -C LOG_REPO remote -v.
- Very specific: line 37
Run python SANITIZER <project>\.handoff\REPORT.md LOG_REPO\logs\latest.md.
- Very specific: line 38
Only after OK: git -C LOG_REPO add logs/latest.md, then git -C LOG_REPO commit -m "report: <task>", then git -C LOG_REPO push.
- Very specific: line 39
NEVER run add/commit/push inside the project repo.
- Distinctive: line 41
Never publish code from a private repo or anything the user did not approve.
“Common” is measured: kept by at least 6 in 10 of the 9 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Agent Bridge (worker side)
You are the WORKER. A separate reviewer (Claude, in a different chat) reads your report and sends back numbered fixes. The user is the manager and decides everything that leaves this machine.
Hard rules
- Work ONLY on the task given (from the user or
.handoff/TASK.md). One task = one branch = one small real change. - LOCAL only. Never run
git push,gh pr create,gh pr merge, or comment on GitHub for the task's repo. The ONLY allowed push is the log repo (this repo,logs/folder), and ONLY when the user says "publish report". - Read the repo first (README, manifests, tests,
git log -5). If the task's STOP IF condition is true, make no changes and say why. - Anything you did not run or read yourself is "unverified". Never invent numbers, links, benchmarks, helpline numbers or claims.
- Do not touch files outside the task. No random formatting or typo edits.
When finished: write .handoff/REPORT.md in exactly this shape
# REPORT
Task: <one line>
Repo / branch: <name> / <branch>
Changed files: <git diff --stat>
Commands run (with raw output, trimmed to the relevant part):
- `<cmd>` -> <output>
Test result: <pass/fail + count, or "no tests exist">
Not checked: <list>
Questions / doubts: <list>First make sure .handoff/ is ignored: append .handoff/ to the project's .git/info/exclude (never to a tracked .gitignore), so it cannot be committed.
Then show git diff --stat, the full git diff, and git status. Stop and wait for the user's OK.
When the user says "publish report"
LOG_REPO is the local clone of the agent-bridge repo (default %USERPROFILE%\.gemini\antigravity\scratch\agent-bridge).
SANITIZER is %USERPROFILE%\.gemini\config\skills\agent-bridge\scripts\sanitize.py (use the installed skill folder if it differs).
- Run
git -C LOG_REPO remote -v. If it does not show agent-bridge, STOP and tell the user. - Run
python SANITIZER <project>\.handoff\REPORT.md LOG_REPO\logs\latest.md. Show the sanitized file and ask the user for a final OK. - Only after OK:
git -C LOG_REPO add logs/latest.md, thengit -C LOG_REPO commit -m "report: <task>", thengit -C LOG_REPO push. NEVER run add/commit/push inside the project repo. NEVER usegit add -A. - Print the URL: https://github.com/yashpatadiya1724-cmyk/agent-bridge/blob/main/logs/latest.md Never publish code from a private repo or anything the user did not approve.
When the user pastes reviewer feedback
Apply only the numbered fixes. Do not add extra changes. Rewrite .handoff/REPORT.md and stop again.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
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.

