agentloop-core
Skill0 starsChanged 4 months agoMITView on GitHub
Master agentloop orchestration for coding and product tasks. Use when the user wants harness mode, full autonomy with gates, or to enter agentloop.
In this repository138 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- .cursor/rules/context-budget.mdcCursor rule
- .cursor/rules/delegation-law.mdcCursor rule
- .cursor/rules/harness-doctrine.mdcCursor rule
- .cursor/rules/verification-law.mdcCursor rule
- agent-interface-designSkill
- agentloop-coreSkill
- agent-loopSkill
- blindspot-passSkill
- brainstorm-prototypesSkill
- change-quizSkill
- conductor-planSkill
- context-auditSkill
- context-compactSkill
- context-surgeonSkill
- contrastingSkill
- deployment-auditSkill
- doctorSkill
- harness-designSkill
- humaniseSkill
- implementation-notesSkill
- implementation-planSkill
- interview-meSkill
- no-ai-slopSkill
- optional-ponytailSkill
- opus-promptingSkill
- persistent-memorySkill
- pev-cycleSkill
- pitch-packagerSkill
- progressive-disclosureSkill
- quality-composeSkill
- reference-huntSkill
- seo-copywritingSkill
- ship-gateSkill
- subagent-swarmSkill
- ui-ux-auditSkill
- warroom-councilSkill
- writer-verifierSkill
- agent-interface-designSkill
- agent-loopSkill
- blindspot-passSkill
- brainstorm-prototypesSkill
- change-quizSkill
- conductor-planSkill
- context-auditSkill
- context-compactSkill
- context-surgeonSkill
- contrastingSkill
- deployment-auditSkill
- doctorSkill
- harness-designSkill
- humaniseSkill
- implementation-notesSkill
- implementation-planSkill
- interview-meSkill
- no-ai-slopSkill
- optional-ponytailSkill
- opus-promptingSkill
- persistent-memorySkill
- The first 60 shown.
The file
We split this skill into the rules it gives an agent
4 rules. About 0 are common, 1 is distinctive, and 3 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
Distinctive · estimate
This author's own choices. Worth a look.
3
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 12
Write goal + done-definition + verify command → .agentloop/goal.md.
- Very specific: line 20
Run ship-gate before claiming done.
- Distinctive: line 62
No Execute before Conductor approval on multi-step work.
- Very specific: line 63
Stop conditions in .agentloop/stop.yaml are hard.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 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
agentloop Core
Workflow
- Cold-start or suspicious env →
/doctor(fix red blockers before spending budget). - Retrieve
.agentloop/handoff.md+.agentloop/memory.md(+ Supermemory if configured). - Write goal + done-definition + verify command →
.agentloop/goal.md. - If the map is fuzzy, discover first (
/unknowns) before spending a build loop. - Conductor Plan (
/plan/conductor-plan) for multi-step work → approval gate. - Choose mode: single | pev | warroom | swarm (smallest first; Overmind rubric).
- Load only the skills you need — compose for quality, don’t spray workers.
- For craft/launch work, prefer
quality-composeoversubagent-swarm. - Execute with narration diet. Keep
implementation-noteswhen working a plan. - Apply
context-surgeoncontinuously; refresh.agentloop/handoff.md. - Run
ship-gatebefore claiming done. Optionallychange-quizbefore merge. - Promote durable facts with
persistent-memory. Final message: outcome first.
Discovery pack (before)
| Need | Skill |
| Unfamiliar area / domain | blindspot-pass |
| Taste you can’t verbalize | brainstorm-prototypes |
| Ambiguity remains | interview-me |
| “Like that” is the spec | reference-hunt |
| Lock decisions (user-facing) | implementation-plan |
| Harness Conductor scaffold | conductor-plan |
| During a planned build | implementation-notes |
| Buy-in doc | pitch-packager |
| Before merge | change-quiz |
| Bloated agent context | context-audit |
| Continuous context budgets | context-surgeon |
| Agent-facing tools | agent-interface-design |
| Oversized skill/spec | progressive-disclosure |
| Setup / connectivity | doctor |
| Long-term memory | persistent-memory |
Craft pack (on demand)
| Need | Skills |
| Kill generic AI output | no-ai-slop |
| Sound human | humanise |
| Personality dial | optional-ponytail (off by default) |
| Screen review | ui-ux-audit |
| Go-live readiness | deployment-audit |
| Public copy | seo-copywriting |
| Contrast (visual + steelman) | contrasting |
| Combine the above | quality-compose |
Guardrails
- Method lenses only — no celebrity impersonation.
- Cap swarm. Prefer one strong pass over five noisy ones.
- External gates beat vibes.
- Finding unknowns is cheaper than rebuilding.
- No Execute before Conductor approval on multi-step work.
- Stop conditions in
.agentloop/stop.yamlare hard.
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.