generated from coulomb/repo-seed
Register WP-0007 (Distribute), WP-0008 (Read-before-Edit), WP-0009 (Measure)
Three workplans queued and registered with the State Hub (via REST — MCP write layer is erroring this session): - AGENTIC-WP-0007 Phase 3 Distribute: per-flavor distributor adapters render approved catalog patterns into proposed (HITL) artifacts, scoped by repo/domain. - AGENTIC-WP-0008 Read-before-Edit reflex: act on the #1 friction finding. - AGENTIC-WP-0009 Phase 4 Measure: baseline + before/after effectiveness + trend. Proceeding in that order. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
This commit is contained in:
100
workplans/AGENTIC-WP-0007-session-memory-phase3.md
Normal file
100
workplans/AGENTIC-WP-0007-session-memory-phase3.md
Normal file
@@ -0,0 +1,100 @@
|
||||
---
|
||||
id: AGENTIC-WP-0007
|
||||
type: workplan
|
||||
title: "Coding Session Memory — Phase 3 (Distribute: per-flavor artifacts, HITL)"
|
||||
domain: helix_forge
|
||||
repo: agentic-resources
|
||||
status: ready
|
||||
owner: codex
|
||||
topic_slug: helix-forge
|
||||
created: "2026-06-07"
|
||||
updated: "2026-06-07"
|
||||
state_hub_workstream_id: "766c9089-d5de-472a-8c0f-85529028cfb9"
|
||||
---
|
||||
|
||||
# Coding Session Memory — Phase 3 (Distribute)
|
||||
|
||||
Implements **Distribute** (PRD §6.4, FR-X1–FR-X4), continuing
|
||||
[AGENTIC-WP-0004](AGENTIC-WP-0004-session-memory-phase2.md) (Curate). Distributor
|
||||
adapters render the **approved / `distribution_ready`** SolutionPatterns from the
|
||||
Pattern Catalog into per-flavor artifacts, using the `rendering_hints` produced in
|
||||
Phase 2. Mirror image of the collector design: **agnostic core, thin adapters at
|
||||
the edges** (FR-A2) — adding a flavor = one collector + one distributor.
|
||||
|
||||
Key boundary (FR-X3): output is **proposed, not auto-applied** — artifacts are
|
||||
written as reviewable proposals (HITL), scoped by repo/domain (FR-X2), with an
|
||||
active-pattern registry tracking which patterns are live where (FR-X4).
|
||||
|
||||
## Distributor Adapter Interface + Artifact Base
|
||||
|
||||
```task
|
||||
id: AGENTIC-WP-0007-T01
|
||||
status: todo
|
||||
priority: high
|
||||
state_hub_task_id: "ff618fa6-a78b-4b80-846b-8cde7ad65451"
|
||||
```
|
||||
|
||||
Define a `Distributor` protocol and an `Artifact` dataclass (flavor, target_path,
|
||||
content, pattern_id) in `session_memory/distribute/`. `render(pattern, scope)`
|
||||
reads the agnostic `SolutionPattern` plus its per-flavor `rendering_hints`; base
|
||||
helpers handle idempotent snippet markers. Agnostic core; flavor logic only in
|
||||
adapters. Unit-tested.
|
||||
|
||||
## Claude Distributor (CLAUDE.md snippet)
|
||||
|
||||
```task
|
||||
id: AGENTIC-WP-0007-T02
|
||||
status: todo
|
||||
priority: high
|
||||
state_hub_task_id: "64f50bd4-1fdf-452e-ae14-890253ab9f33"
|
||||
```
|
||||
|
||||
`distribute/claude.py`: render an approved pattern into a `CLAUDE.md` snippet block
|
||||
(or skill stub) with stable `BEGIN/END` markers so re-distribution updates in
|
||||
place rather than duplicating. Uses `rendering_hints["claude"]`. Unit-tested.
|
||||
|
||||
## Codex + Grok Distributors
|
||||
|
||||
```task
|
||||
id: AGENTIC-WP-0007-T03
|
||||
status: todo
|
||||
priority: high
|
||||
state_hub_task_id: "382790f5-1fb4-4394-b039-1649cbf3b20a"
|
||||
```
|
||||
|
||||
`distribute/codex.py` (`AGENTS.md` snippet) and `distribute/grok.py` (native
|
||||
instruction format), each rendering the *same* agnostic pattern via its
|
||||
`rendering_hints`. Confirms FR-A3: a pattern discovered via one flavor is
|
||||
expressible for all. Unit-tested.
|
||||
|
||||
## Scoping + Proposed-Not-Applied Output + Active-Pattern Registry
|
||||
|
||||
```task
|
||||
id: AGENTIC-WP-0007-T04
|
||||
status: todo
|
||||
priority: high
|
||||
state_hub_task_id: "2c690f29-2aee-460a-b9cd-3566018f6b3c"
|
||||
```
|
||||
|
||||
Filter patterns by `Scope` (repos/domains/flavors) so a pattern only lands where it
|
||||
applies (FR-X2). Write artifacts as **proposals** under a `proposals/` dir, never
|
||||
auto-applied (FR-X3, HITL). Track which patterns are active in which environments
|
||||
in an active-pattern registry (FR-X4). Unit-tested.
|
||||
|
||||
## Distribute Entrypoint + Tests + Verify
|
||||
|
||||
```task
|
||||
id: AGENTIC-WP-0007-T05
|
||||
status: todo
|
||||
priority: medium
|
||||
state_hub_task_id: "f9e24c13-7049-4c1c-a2d6-3a4dc4e752fd"
|
||||
```
|
||||
|
||||
`python -m session_memory.distribute`: read approved catalog patterns, render
|
||||
per-flavor proposals scoped by repo/domain, emit a proposal summary + JSON.
|
||||
Document in `session_memory/README.md`. Verify end-to-end against the real catalog.
|
||||
After workplan updates, notify the operator to run from `~/state-hub`:
|
||||
|
||||
```bash
|
||||
make fix-consistency REPO=agentic-resources
|
||||
```
|
||||
56
workplans/AGENTIC-WP-0008-read-before-edit-reflex.md
Normal file
56
workplans/AGENTIC-WP-0008-read-before-edit-reflex.md
Normal file
@@ -0,0 +1,56 @@
|
||||
---
|
||||
id: AGENTIC-WP-0008
|
||||
type: workplan
|
||||
title: "Act on #1 friction — Read-before-Edit reflex"
|
||||
domain: helix_forge
|
||||
repo: agentic-resources
|
||||
status: ready
|
||||
owner: codex
|
||||
topic_slug: helix-forge
|
||||
created: "2026-06-07"
|
||||
updated: "2026-06-07"
|
||||
state_hub_workstream_id: "6aac5cfc-4799-4d07-9537-42a203af2d1b"
|
||||
---
|
||||
|
||||
# Act on #1 Friction — Read-before-Edit Reflex
|
||||
|
||||
The error-body mining ([AGENTIC-WP-0006](AGENTIC-WP-0006-error-body-mining.md))
|
||||
found that the single most common error across real coding sessions is
|
||||
**`File has not been read yet. Read it first before writing to it.`** — Edit/Write
|
||||
before Read, in **12 of 27 sessions across 8 repos** — followed by the stale-read
|
||||
**`File has been modified since read`** (6 sessions). See
|
||||
[ASSESSMENT-infra-friction.md](../docs/ASSESSMENT-infra-friction.md).
|
||||
|
||||
This is the cheapest high-value fix surfaced by the whole analysis: a short
|
||||
behavioural reflex in the agent instructions. We also capture it as a curated
|
||||
SolutionPattern so Phase 3 Distribute can propose it to other repos/flavors —
|
||||
closing the assess → curate → distribute loop by hand for one real pattern.
|
||||
|
||||
## Add Read-before-Edit Reflex to Agent Instructions
|
||||
|
||||
```task
|
||||
id: AGENTIC-WP-0008-T01
|
||||
status: todo
|
||||
priority: high
|
||||
state_hub_task_id: "549c84c1-5bd8-4ff6-b61d-1c72946b8b8e"
|
||||
```
|
||||
|
||||
Add a concise, data-cited **Read-before-Edit / re-read-on-"modified since read"**
|
||||
reflex to `AGENTS.md` (and note for `CLAUDE.md`), targeting the #1 and #2 recurring
|
||||
errors. Keep it short to avoid context bloat (cf. PRD OQ6 — pattern bloat degrades
|
||||
context budgets).
|
||||
|
||||
## Capture as Curated SolutionPattern for Distribute
|
||||
|
||||
```task
|
||||
id: AGENTIC-WP-0008-T02
|
||||
status: todo
|
||||
priority: medium
|
||||
state_hub_task_id: "c007baf9-db14-40fa-b944-d1f1a71ea28b"
|
||||
```
|
||||
|
||||
Promote the recurring "file not read" problem into a curated `SolutionPattern` in
|
||||
the Pattern Catalog with per-flavor `rendering_hints`, so Phase 3 Distribute can
|
||||
render and propose it across repos/flavors. Links assess → curate → distribute end
|
||||
to end on a real pattern. After updates, notify the operator to run
|
||||
`make fix-consistency REPO=agentic-resources`.
|
||||
68
workplans/AGENTIC-WP-0009-session-memory-phase4.md
Normal file
68
workplans/AGENTIC-WP-0009-session-memory-phase4.md
Normal file
@@ -0,0 +1,68 @@
|
||||
---
|
||||
id: AGENTIC-WP-0009
|
||||
type: workplan
|
||||
title: "Coding Session Memory — Phase 4 (Measure: effectiveness + fleet trend)"
|
||||
domain: helix_forge
|
||||
repo: agentic-resources
|
||||
status: ready
|
||||
owner: codex
|
||||
topic_slug: helix-forge
|
||||
created: "2026-06-07"
|
||||
updated: "2026-06-07"
|
||||
state_hub_workstream_id: "99f1d836-3be0-40e5-9f17-63d3ecc5fcca"
|
||||
---
|
||||
|
||||
# Coding Session Memory — Phase 4 (Measure)
|
||||
|
||||
Implements **Measure** (PRD §6.5, FR-M1–FR-M3) — the loop-closer. After patterns
|
||||
are distributed (Phase 3) and changes land (e.g. the State Hub skill
|
||||
[STATE-WP-0058] and the Read-before-Edit reflex
|
||||
[AGENTIC-WP-0008](AGENTIC-WP-0008-read-before-edit-reflex.md)), Measure answers:
|
||||
**did it actually help?**
|
||||
|
||||
Reuses what is already captured — WP-0005 tool buckets, WP-0006 error mining — so
|
||||
this is computation over existing digests, not new capture.
|
||||
|
||||
## Baseline Metrics Module + Persisted Baseline
|
||||
|
||||
```task
|
||||
id: AGENTIC-WP-0009-T01
|
||||
status: todo
|
||||
priority: high
|
||||
state_hub_task_id: "e5c2016a-2d51-4382-a013-7153e053e8ed"
|
||||
```
|
||||
|
||||
`session_memory/measure/metrics.py`: compute fleet metrics over real sessions
|
||||
(infra-overhead share, error rate, recurring-error count, schema-thrash, cost
|
||||
percentiles) and persist a **timestamped baseline snapshot**. Reuses
|
||||
`detect.signals.tool_bucket` and the digest `error_snippets`. Unit-tested.
|
||||
|
||||
## Before/After Per-Pattern Effectiveness
|
||||
|
||||
```task
|
||||
id: AGENTIC-WP-0009-T02
|
||||
status: todo
|
||||
priority: high
|
||||
state_hub_task_id: "aa097a00-3462-41da-a137-67e1d61d8d33"
|
||||
```
|
||||
|
||||
Given a change/pattern with an applied-at date, compare sessions **after** it
|
||||
against the pre-change baseline (cost, error rate, infra-overhead, success) to
|
||||
surface **per-pattern effectiveness** so ineffective patterns can be revised or
|
||||
retired (FR-M1/FR-M2). Unit-tested.
|
||||
|
||||
## Fleet-Trend Report + Entrypoint + Tests
|
||||
|
||||
```task
|
||||
id: AGENTIC-WP-0009-T03
|
||||
status: todo
|
||||
priority: medium
|
||||
state_hub_task_id: "f1147d59-2fb7-4d35-baec-b8f001bb9d62"
|
||||
```
|
||||
|
||||
`python -m session_memory.measure`: fleet-level trend (is the median session
|
||||
getting cheaper / more reliable over time, FR-M3) plus per-pattern effectiveness;
|
||||
markdown + JSON. Document in `session_memory/README.md`. After updates, notify the
|
||||
operator to run `make fix-consistency REPO=agentic-resources`.
|
||||
|
||||
[STATE-WP-0058]: handed off to the state-hub repo worker
|
||||
Reference in New Issue
Block a user