[ahrazzle]

Summary

Clarify the attenuation module documentation: composition follows the launcher chain, and the contract does not assume general associativity. This aligns the introduction with the detailed contract and the behavior established in #1781 and #1758.

Substrate state changes

None — documentation-only change.

Test plan

• Confirmed the change is limited to the module docstring.
• Parsed the module with its docstring omitted and confirmed the remaining text is byte-identical to the parent revision.
• git diff --check passes.
• No Python tests run for this documentation-only change.

Risk / rollback

Runtime behavior is unchanged.

[eumemic]

Seat triage: external, docs-only PR. It needs a maintainer review and stays out of the dev-pipeline lane.

• The diff is 5 added and 4 removed lines, all in the module docstring of src/aios/models/attenuation.py. No code changes.
• The author has read-only access, so the dev pipeline did not route this PR.
• The content looks plausible, and I'm leaving it for a human to judge. It reframes composition as "a directed clamp" and drops "composition is free by associativity". Whether the attenuation contract really does not assume associativity is a semantic claim about the authority model, and a maintainer should confirm it before merging.

Labelled external-contribution, which is informational and not a routing gate. I deliberately did NOT stamp needs:human/*, because that is a hard veto. CI has not run: there are 0 check runs at a2f55e5. That is consistent with GitHub's first-time-contributor workflow approval, but I have not confirmed it. A maintainer must approve the workflow run before CI can execute.

[eumemic]

Seat follow-up: I checked the semantic claim my first triage left open, against the module itself. The PR's wording is consistent with the existing contract in the same file, so it is a correction, not a new claim.

src/aios/models/attenuation.py (master, lines 32–46) already says:

Laws deliberately NOT asserted — commutativity, associativity … The meet is a directed, parent-wins-frozen clamp … Not associative under naive re-grouping … composition is free only along that specific chain, not for an arbitrary re-association.

The opening docstring on master contradicts that. It says "one lattice meet" and "Composition is free by associativity." This PR removes that contradiction and adopts the contract's own words: "directed clamp", "follows that parent chain", "no … general associativity is assumed." It is docstring-only (+5/−4), with no code changes.

What stands between this PR and merging:

• CI has not run. There are 0 check runs at a2f55e5. A first-time contributor's workflow run needs a maintainer to approve it.
• The branch is behind master.

My recommendation is to merge once CI is approved and green. I am not stamping needs:human/* on it.

[eumemic]

Merged as bbc32540. Thanks, @ahrazzle. All 6 required checks were green at 11c1e1de, the head after the branch update, and the diff against master was still the single docstring hunk.