mirror of
https://github.com/boshu2/agentops.git
synced 2026-09-14 15:08:13 +08:00
docs(loop): fold agile re-plan into the spine — re-plan on evidence, not just failure (age-az5 #spine-replan)
Completes the doctrine propagation (age-hx7 → age-ni4 → here). The spine had
principle 7 ('the map is fixed; the route is re-routed... recalculated on
failure'); this adds the move-7→move-1 closing subsection that sharpens it:
re-routing fires on EVIDENCE (a successful wave's evidence can also
refactor/insert/drop/reorder the remaining waves), not only on failure. Names
the retry-not-replan + waterfall anti-patterns; links the full mechanics in
skills/rpi/references/agile-replan-loop.md. Extends principle 7, does not
duplicate it. Added as a new subsection (not a rewrite of the move-7 paragraph)
to avoid colliding with the in-flight move-7 edit.
Closes-scenario: age-az5#spine-replan-on-evidence
Bounded-context: BC3-Loop
Evidence: docs/architecture/operating-loop.md (### The loop closes here)
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
This commit is contained in:
@@ -121,6 +121,10 @@ example to the test, gate, or eval that proved it.
|
||||
|
||||
Two outputs per loop turn — evidence into `.agents/ratchet/` and the bead; learnings only if they cleared the promotion bar (next section). Skills: `/post-mortem`, `/forge`, `/retro`, `/ratchet`, `/flywheel`, `/harvest`.
|
||||
|
||||
### The loop closes here: re-plan on evidence, not just on failure
|
||||
|
||||
Move 7 feeds **back into move 1** — this is where the route gets re-routed (principle 7). The sharpening principle 7 leaves implicit: re-routing is triggered by **evidence, not only by failure.** A wave that *succeeds* still teaches something the plan didn't know, and that evidence may **refactor, insert, drop, reorder, or re-scope the *remaining* waves** before the next one runs. The wave plan is a hypothesis; each wave is the experiment that tests it. Under `--auto` the orchestrator executes those pivots itself — it is not gated on a wave *failing* first, and it does not run the initial wave-list to the letter. Two failure modes this kills: **retry-not-replan** (re-cranking a failed wave forever instead of asking whether the *remaining plan* should change) and **waterfall** (executing the pre-written wave list because "that was the plan"). Bounded by the run's circuit breakers (budget / attempt cap / oscillation detection) and the ≥5-arc post-mortem checkpoint; the operator is surfaced only at the terminal objective or a breaker trip — never just to approve a pivot. The orchestrator that owns this across a turn is `/rpi`; full mechanics: [Agile Re-Plan Loop](../../skills/rpi/references/agile-replan-loop.md).
|
||||
|
||||
## The promotion ratchet
|
||||
|
||||
Do not run full ceremony for every observation. Promote progressively:
|
||||
|
||||
Reference in New Issue
Block a user