Skip to content

dankofly/perfectify / skill/dagx-agi-kernel

dagx-agi-kernel

Verified

Improve and verify agent work after repeated failures, in dependency-heavy tasks, or when optimization claims need baseline and regression evidence. Use for DAGx/Perfectify requests, failed retries, risky multi-step work, or requests to verify an improvement. Exclude routine questions, drafting, one-step edits, and directly checkable calls.

templates/trial-ledger.md

View on GitHub
# Trial Ledger

Use one ledger for one baseline/challenger evaluation. Record observed values only. Use `null` when a value was not measured.

## Experiment Contract

| Field | Value |
| --- | --- |
| Evaluation ID | `<id>` |
| Goal and target gap | `<observable gap>` |
| Baseline version | `<version or artifact hash>` |
| Challenger version | `<version or artifact hash>` |
| Model and version | `<exact identifier>` |
| Harness and configuration | `<version and material settings>` |
| Cases and held-out rule | `<dataset version and split rule>` |
| Run order or randomization | `<method>` |
| Acceptance rule | `<ordered must-pass conditions>` |
| Protected behavior | `<regression gates>` |
| Resource contract | `<attempt, token, tool, time, compute, and cost limits>` |
| Evaluator | `<deterministic check or blinded adjudicator>` |

## Trial Records

| Case | Condition | Attempt | Activated | Success | Acceptance evidence | Protected failures | Input tokens | Output tokens | Tool calls | Wall time ms | Cost | Artifact or trace | Notes |
| --- | --- | ---: | --- | --- | --- | ---: | ---: | ---: | ---: | ---: | ---: | --- | --- |
| `<case-id>` | `baseline` or `candidate` | 1 | `true`, `false`, or `null` | `true`, `false`, or `null` | `<verifier result>` | 0 | `null` | `null` | `null` | `null` | `null` | `<URI or hash>` | `<material limitation>` |

## Promotion Record

| Decision field | Result |
| --- | --- |
| Activation precision and recall | `<separate values plus denominators>` |
| Paired success delta | `<percentage points plus paired count>` |
| Paired token delta | `<absolute and percent plus paired count>` |
| Protected failures | `<baseline and candidate counts>` |
| Held-out result | `<coverage and result>` |
| Complexity or context delta | `<measured change>` |
| Decision | `promote`, `preserve`, `rollback`, or `insufficient evidence` |
| Decisive evidence | `<artifact, trace, or result hash>` |
| Residual uncertainty | `<unevaluated scope>` |

Do not promote from this template alone. The recorded evidence and evaluation design determine whether an improvement claim is supported.
reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis.Stars: 471kUpdated Claude CodeCodexVerifiedInstall
diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness.Stars: 322kUpdated Claude CodeCodexVerifiedInstall
dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependenciesStars: 322kUpdated Claude CodeCodexVerifiedInstall
executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is availableStars: 322kUpdated Claude CodeCodexVerifiedInstall