precision
Skill2 starsChanged 8 months agoMITView on GitHub
Deliberate verification for high-stakes tasks.
In this repository18 other files it gives its agents.
- auditingSkill
- bootstrappingSkill
- debuggingSkill
- executingSkill
- finalizingSkill
- guardingSkill
- handling-reviewSkill
- ideatingSkill
- isolatingSkill
- orchestratingSkill
- parallelizingSkill
- planningSkill
- requesting-reviewSkill
- runbookSkill
- scaffoldingSkill
- skill-craftingSkill
- tddSkill
- verifyingSkill
The file
We split this skill into the rules it gives an agent
1 rule. About 0 are common, 1 is distinctive, and 0 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.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 10
Before producing any output, apply these checks:
≈ “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
Calibrating for Precision
Protocol
Before producing any output, apply these checks:
- Decompose — Break the task into atomic claims or steps. Evaluate each independently.
- Trace assumptions — Identify every assumption. Verify each against available evidence. Flag unknowns explicitly rather than interpolating.
- Cross-check — For calculations, derive the result twice using different methods. For code, mentally execute the path. For facts, confirm source reliability.
- State confidence — When certainty is partial, say so. "Uncertain" is more useful than plausible-sounding fabrication.
- Slow the final pass — Re-read the complete output before delivering. Check for contradictions, off-by-one errors, stale references, and copy-paste artifacts.
Scope
This applies to every response in the session. No output is trivial enough to exempt.
- Code: step through logic, check edge cases, verify types.
- Math: re-derive before committing. Show intermediate steps when non-obvious.
- Facts: assert only what is confirmed. Qualify the rest.
- Shell commands: trace the effect of flags and arguments before execution.
Gotchas
- Confidence masking ignorance — Fluent phrasing feels correct even when wrong. Pause on any claim that "sounds right" but lacks traced reasoning.
- Anchoring on first result — The initial answer biases subsequent checks. Derive independently, then compare.
- Precision theater — Adding caveats without actually rechecking is cosmetic, not calibration. Do the work.
- Rushing the last 10% — Most errors land in final assembly: wrong variable name, stale value, inverted condition. The last pass matters most.
- Skipping verification on "simple" tasks — Trivial arithmetic and one-liner code changes produce a disproportionate share of shipped bugs.
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.

