golem15/msd-core
4
0
Fork 0
Code Issues 1 Pull Requests Actions 2 Packages Projects Releases Wiki Activity

feat(#2632): executor actuals and the closed estimate-calibration loop (#2672)

Browse Source 
* feat(#2632): record executor actuals and close the estimate calibration loop

* fix(#2632): calibrate against the raw projection so the loop converges

* test(#2632): add closed-loop convergence guard and codify the feedback-loop rule

* fix(#2632): pair calibration samples per plan; atomic write; amend adr

* chore(#2632): backfill changeset pr to 2672

* fix(#2632): retry renameSync on transient windows errnos and clean up the temp
This commit is contained in:
Tom Boucher
2026-07-26 16:28:56 -04:00
committed by GitHub
parent 920a5f3f06
commit bd570618d4
37 changed files with 943 additions and 147 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
docs/reference/plan-md.md
View File

@@ -73,7 +73,7 @@ must_haves:
| `requirements` | Yes | array of IDs | Requirement IDs from ROADMAP.md that this plan addresses. Every phase requirement ID must appear in at least one plan's `requirements` field. Empty arrays are a BLOCKER. |
| `user_setup` | No | array of objects | External-service setup steps that Claude cannot automate (account creation, secret retrieval, dashboard configuration). When present, execute-phase generates a `USER-SETUP.md` checklist for the developer. |
| `status` | No | `superseded` | Marks a plan that was deliberately reassigned or abandoned mid-phase and will never be executed. A `status: superseded` plan is excluded from the phase's plan and summary counts, so it never holds the phase below 100%. See [Superseded plans](#superseded-plans). Any other value (or the field's absence) has no effect on counting. |
| `estimate` | No | object | Projected execution cost: `{tokens, tasks, confidence}` (#2631, [ADR-2629](../adr/2629-phase-effort-estimation-calibration.md)). `tokens` is an `estimateTokens`-scale projection with the project's calibration factor **already applied** (which is why the plan-checker passes `--calibrated` to `estimate-check` — re-applying it would square the correction); `confidence` (`low`/`med`/`high`) is **derived from the calibration sample count, never self-rated**. Additive and optional — a plan without it behaves exactly as before. A plan estimated above `workflow.smart_zone_tokens` is flagged with a split recommendation at plan time; the flag is advisory and never blocks. |
| `estimate` | No | object | Projected execution cost: `{tokens, raw_tokens, tasks, confidence}` (#2631, [ADR-2629](../adr/2629-phase-effort-estimation-calibration.md)). `tokens` is an `estimateTokens`-scale projection with the project's calibration factor **already applied** (which is why the plan-checker passes `--calibrated` to `estimate-check` — re-applying it would square the correction); `confidence` (`low`/`med`/`high`) is **derived from the calibration sample count, never self-rated**. Additive and optional — a plan without it behaves exactly as before. A plan estimated above `workflow.smart_zone_tokens` is flagged with a split recommendation at plan time; the flag is advisory and never blocks. |
| `must_haves` | Yes | object | Goal-backward verification criteria. See below. |
### Superseded plans

11
docs/reference/planning-artifacts.md
View File

@@ -189,6 +189,17 @@ See [PLAN.md schema](plan-md.md) for the full field reference.
| **Produced by** | `execute-phase` executor agent (written at the end of each plan's execution). |
| **Consumed by** | `/gsd-progress` (phase status); `gsd-planner` (when a subsequent plan has a genuine dependency on prior plan output); `milestone-summary`. |
**`actuals` frontmatter (#2632, [ADR-2629](../adr/2629-phase-effort-estimation-calibration.md)).** When the phase's PLAN carried an `estimate`, the executor records what it actually cost:
```yaml
actuals:
tokens: 74000 # estimateTokens scale (chars/4) over the realized diff
tasks: 5
commits: 7
```
`tokens` uses the **same scale as the estimate**, not a harness-reported token count — an executor subagent cannot read its own consumption, and a ratio between two different measurement methods would measure the methods rather than the miss. `/gsd:extract-learnings` pairs each phase's estimate with its actuals via `gsd_run query estimate-calibrate`, writes `.planning/estimation-calibration.json`, and the planner applies the resulting factor to subsequent estimates. Additive and optional: a summary without `actuals` simply contributes no calibration sample.
### `<NN>-VERIFICATION.md`
| | |