peptide-repo/prautoblogger
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat(runs): state machine, idempotent resume, reaper extension, retention

Browse Source 
Pipeline v2 Phase 1, commit 4/4. Run_Stage_State persists per-run
per-stage state keyed run_id + stage + agent_role + item_key (item_key
scopes articles because one run_id spans a whole batch; agent_role is
the Phase-2 quorum dimension, schema-only). Done stages snapshot output
and are reused on re-entry — writer stages via
Content_Generator::set_run_item, the review via its stored verdict, a
published item skipped outright — never re-run, never re-charged. Post
creation is keyed by _prautoblogger_run_id + new _prautoblogger_idea_hash
(check-before-insert in Publisher) so retries cannot duplicate posts.
Run-level research/analysis checkpoints + done/failed lifecycle marks
ride Pipeline_Runner/Executor (incl. orphan-abort + budget-abort paths).

Run_Reaper rides the EXISTING #19 daily cron: stages stuck running >2x
expected stage wall-clock and runs silent >2x expected run wall-clock ->
failed (filterable), open stages reaped, queryable as 'incomplete'.
Retention: request_json + stage output payloads NULLed after R days, R
from the new Audit Payload Retention setting (default constant 14,
0 = keep forever, no literal at point of use). Audit_Writer records
editorial verdicts (run_decisions) and per-article kept sources
(run_sources) for new runs.

Under-300: status-transient/summary helpers extracted verbatim to
Pipeline_Status (Pipeline_Runner 271; removed the worker's duplicate).

Agent-Session: pipeline-v2-phase1-engineer-20260612
This commit is contained in:
peptiderepo
2026-06-11 04:31:55 +00:00
parent 04f09f5f24
commit ed8ec9425f
17 changed files with 1358 additions and 102 deletions
Download Patch File Download Diff File Expand all files Collapse all files

17
CONVENTIONS.md
View File

@@ -204,13 +204,18 @@ the uninstall prefix-sweep effective), then cover it in the unit ladder tests.
## How To: Add a New Pipeline Stage
The writing pipeline is an ordered array of stages. Each stage is a class method on `Content_Generator`.
Writing stages are methods on `Content_Generator` that delegate to its `execute_stage()`
helper (Opik span + LLM dispatch + cost log + run-stage state in one place).
1. Add the stage method to `class-content-generator.php`: `protected function stage_{name}(PRAutoBlogger_Content_Request $request): string`
2. Register the stage in the `get_pipeline_stages()` method for the appropriate mode (single_pass or multi_step)
3. Each stage receives the accumulated context and returns its output
4. Each stage logs its cost via `Cost_Tracker::log_api_call()`
5. If a stage fails after retries, the entire pipeline fails (no partial publishes)
1. Add the stage method to `class-content-generator.php` and call
`$this->execute_stage( $stage, $span_name, $user_prompt, $request, $model, $options )`
2. Wire it into `generate_multi_step()` (or the relevant mode path)
3. Add the stage to `PRAutoBlogger_Stage_Display_Map` (label, default agent role, prompt key)
and — if it has registry-managed copy — a prompt key in the defaults classes
4. Cost logging and per-run stage state (idempotent resume, output snapshot) come free from
`execute_stage()`; the per-run cost governor guards the call inside the provider
5. If a stage fails after retries, the entire pipeline fails (no partial publishes); the
stage row is failed by the worker's catch-all and the run is swept by Run_Reaper
---