docs(positioning): engineer context carries the ADR-0011 line (abilityai/trinity-enterprise#800)

[vybe]

Fixes abilityai/trinity-enterprise#800 — the engineer-context half of the ADR-0011 positioning epic (abilityai/trinity-enterprise#796). Independent branch off dev — the four #796 children touch disjoint files and merge in any order (operator, 2026-10-05). Draft until the chain is in.

What

Nine files, copy only, in three commits:

• Engineer context + template card (0d1a2dc1): CLAUDE.md Project Overview line and the :335 heading; AGENTS.md opening line; template.yaml description (T6 — headline sentence, then "This agent develops it: it implements, reviews and ships changes to the abilityai/trinity repository."), display_name unchanged (T5).
• Memory, onboarding and diagram docs (348e4263): docs/memory/requirements.md, docs/memory/requirements/overview.md, docs/memory/architecture.md (title + overview paragraph), docs/onboarding/00-welcome.md (tagline, "What is Trinity?" lead, and the "Learn from experience" bullet deleted — T1), docs/diagrams/01-system-overview.md (intro, lead, and the 83-column box line), docs/trinity-architecture-diagrams.md (**Platform**: label — T3; ### 1. Agent Orchestration — T4).
• Fix after review (a7cf6707): the supporting-facts clause names all three shipped runtimes — "model-agnostic: each agent runs on Claude Code, Gemini CLI or OpenAI Codex" — in CLAUDE.md, docs/memory/architecture.md, docs/memory/requirements/overview.md, docs/diagrams/01-system-overview.md; the same list in AGENTS.md and the welcome page's multi-runtime bullet.

The issue's acceptance grep returns nothing on the branch (14 hits at the base — the 13 table rows plus docs/trinity-architecture-diagrams.md:570 — 0 at the tip).

Rulings carried (orchestrator, on the operator's behalf — plan file)

• T1 00-welcome.md "📊 Learn from experience — Build semantic memory over time": deleted (planner's recommendation) — a forbidden claim under the epic's rules, and it duplicated "🧠 Remember persistently" two lines above. One bullet inside a table file, not a section removal.
• T2 01-system-overview.md "the Trinity autonomous agent platform" → "the Trinity platform".
• T3 trinity-architecture-diagrams.md label **Platform Version**: → **Platform**:.
• T4 trinity-architecture-diagrams.md ### 1. Autonomous Agent Orchestration → ### 1. Agent Orchestration — needed for the acceptance grep to reach zero; a mechanism heading, no claim. The Process Engine line below it stays with trinity#2491.
• T5 template.yaml display_name stays Trinity Platform — it carries no old claim; the issue quotes the description as the problem. Rename is the operator's call.
• T6 template.yaml description is the hybrid — headline sentence first, then what this agent does — so the fleet map and Library card never show the dev agent as if it were the product.
• F1 (after review): the runtime list names all three shipped runtimes (docs/memory/architecture/agent-runtime.md "Three ship today"); AGENTS.md reads "plain Claude Code, Gemini CLI or OpenAI Codex projects". Only lines already rewritten by this branch plus the welcome page's multi-runtime bullet; the welcome page's runtime list further down (no Codex entry) is reported, not restructured.

Review + security

/review (claude-fable-5-1, report-only): MERGEABLE AFTER FIXES — 0 critical. Exactly 9 files vs the plan's final reviewed text; welcome bullet deleted cleanly; the 83-column box aligned; template.yaml parses, one key changed, no test reads it; .claude, src/backend/enterprise untouched and the root CLAUDE.md change is exactly the ruled lines (no skill-injection text); git merge-tree against live dev (493d2a6) clean. I1 (informational, the dropped third runtime) fixed in a7cf6707. I2 (record only): the description is one physical line inside the | block — the faithful form. /cso --diff: 0 findings.

Tests

• tests/unit/test_2339_testing_docs_consolidated.py + tests/unit/test_2306_architecture_split.py: 42 passed at 348e4263 and again after the fix. The enterprise-docs-guard pattern: 0 hits over the changed files. template.yaml display_name still 'Trinity Platform'.
• No unit test pins any old wording and none reads the root template.yaml. Not run here: the full unit island — CI on this PR.

Before merge

• DoD bullet 2, second half — "the next /discover-agents run shows the new description in fleet/system-map.yaml": trinity-pm runs it after this merges and the engineer's workspace syncs; the Library card picks the description up from template.yaml on its own.
• display_name kept as Trinity Platform (T5) — say the word if the card should be retitled.
• No open PR touches any of the 9 files (23 open PRs scanned 2026-10-05 23:40 UTC); merge-tree against dev @ 493d2a6 clean.

Handoffs

• Abilityai/trinity-dev agents/trinity-code-analyzer.md:14 ("an autonomous agent orchestration and infrastructure platform") — lives in the .claude submodule, which this branch never stages; one-line change through trinity-dev's own flow (comment on the issue).
• Pre-existing two-runtime wording outside this issue's table, not edited: docs/memory/feature-flows/agent-to-agent-collaboration.md:24, docs/trinity-architecture-diagrams.md:612, docs/diagrams/03-agent-container.md:7, docs/onboarding/01-getting-started.md:227, the 00-welcome.md runtime list (no Codex entry); src/backend/models.py:226,313 stale "claude-code" or "gemini-cli" comment. docs/archive/* hits are history.
• The planner's proposal for a CI guard against the retired phrases is a new capability — posted on the issue, not built here.
• abilityai/trinity-enterprise#796 — epic. Siblings: fix(executor): salvage telemetry + auto-retry reader-race empty results (#678) #797 (docs(positioning): README, distribution surfaces and first-run screen carry the ADR-0011 line (abilityai/trinity-enterprise#797) #3258), fix(circuit-breaker): only count TCP unreachability toward circuit (#474) #798 (docs(positioning): user-docs 'What is Trinity' leads with the operating-system line (abilityai/trinity-enterprise#798) #3257), chore(tooling): mark root pyproject as uv-unmanaged #801 (feat(positioning): runtime prompts carry the Trinity one-liner (abilityai/trinity-enterprise#801) #3259).

🤖 Generated with Claude Code

[trinity-ability]

merge-train: batch validated on train/20261006-1038 (#3270)