outcome-lock
Skill1 starsChanged 2 months agoMITView on GitHub
Preserve a principal's original user-visible outcome across long autonomous runs, plans, Lenny conductors, councils, scope cuts, safety gates, and completion claims. Use when locking a goal, ratifying a plan, running Garcia/Lenny, convening a terminus council, or preventing semantic goal drift. A blocker may pause delivery but may never redefine “done.”
In this repository20 other files it gives its agents.
- AGENTS.mdAGENTS.md
- bayesianSkill
- claudes-gambleSkill
- codebase-designSkill
- councilSkill
- elon-first-principlesSkill
- find-skillsSkill
- first-principlesSkill
- fleet-conductorSkill
- garciaSkill
- gbrain-captureSkill
- grill-meSkill
- karpathySkill
- kissingerSkill
- meta-conductorSkill
- pre-kSkill
- release-managerSkill
- rumsfeldSkill
- setup-lennySkill
- ship-conductorSkill
The file
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
- could be automatic: a setting could enforce it
- Very specific: line 12
Write an OUTCOME-CONTRACT.json beside the plan or run ledger:
- Distinctive: line 50
Do not substitute “merge-ready,” “within authorized scope,” “default-off,” “held,” “simulated” or “documented” for a broader runnable …
- Distinctive: line 69
If scope changed without principal approval: stop the narrowed path, restore the original finish line, mark any honest blocker and continue …
- Very specific: line 75
Every council prompt starts with the verbatim originalOutcome and a criterion-by-criterion table.
≈ “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
Outcome Lock
Lock what the principal asked users to be able to do, not merely the artifact an agent proposes to build.
Create the contract before work
Write an OUTCOME-CONTRACT.json beside the plan or run ledger:
{
"version": 1,
"originalOutcome": "Verbatim user-visible outcome",
"lockHash": "",
"status": "in_progress",
"criteria": [
{"id": "C1", "text": "Externally observable success", "required": true, "status": "pending", "evidence": []}
],
"blockers": [],
"scopeChanges": []
}Stamp it once, then validate it at checkpoints and before any completion claim:
node <this-skill>/scripts/validate-outcome.mjs OUTCOME-CONTRACT.json --stamp
node <this-skill>/scripts/validate-outcome.mjs OUTCOME-CONTRACT.json
node <this-skill>/scripts/validate-outcome.mjs OUTCOME-CONTRACT.json --claim-completePrincipal-owned invariant
This outcome may not be narrowed, reinterpreted or replaced by an agent, council, release cutoff, safety gate or evidence manifest. Any proposed change requires the principal to approve the exact before-and-after wording.
- Preserve
originalOutcomeand every required criterion verbatim after stamping. - Agents may propose a scope change but may not approve it or restamp the contract. Record the proposal and continue all unaffected work. Only an explicit principal response quoting or unambiguously accepting the before/after wording authorizes a new contract version.
- A hard rail, missing credential, production approval, financial canary or
provider limitation changes the status to
blocked; it never changes the outcome and never permitscomplete. - Do not substitute “merge-ready,” “within authorized scope,” “default-off,” “held,” “simulated” or “documented” for a broader runnable outcome unless the original contract says that is the outcome.
Automatic checkpoint
At each phase boundary and every 30–60 minutes, compare actual work to the locked criteria. The user does not need to monitor this.
Original outcome:
Criteria passed: X/Y
Observable progress:
Current blocker:
Next action:
Was scope changed? YES/NO
Reorientation performed (if YES):If scope changed without principal approval: stop the narrowed path, restore the original finish line, mark any honest blocker and continue the next unaffected action. Never merely report drift and proceed with it.
Council rule
Every council prompt starts with the verbatim originalOutcome and a
criterion-by-criterion table. Each seat must mark every required criterion
PASS, FAIL, BLOCKED or UNPROVEN with evidence. Councils may critique
implementation or propose a principal decision; they may not rewrite the
outcome.
The terminal question is the original user-visible question, not “is this diff good?” A terminus GO requires every required criterion to pass. Otherwise the verdict is NO-GO or BLOCKED and the run remains open.
Evidence honesty
Match evidence to the claim. Unit tests, fixtures, local QA, read-only provider calls, financial canaries, production UAT and ordinary-user UAT are different proof classes. Narrower evidence cannot satisfy a broader criterion.
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.

