docs(corpus): standard for taxonomy (#1324)

[serina-mcfall]

Summary

Adds launchpad/docs/corpus/standards/taxonomy.md, the canonical policy node for the type front-matter field: where its 13 closed-enum values originate, how to choose among them when a node's subject could plausibly fit more than one, and what to do if none of the 13 fit. Links to COMPATIBILITY.md for the enum-addition procedure rather than restating it.

Related issue

Closes #1324

Issue type

Task

---

Agent provenance

Field	Value
Harness / provider	Claude Code
Model	claude-sonnet-5
Session reference	N/A
Initiating human	N/A

Objective

Create the one hand-authored canonical corpus document launchpad/docs/corpus/standards/taxonomy.md, governing the type field's 13-value closed enum.

Impacted components

• launchpad/docs/corpus/standards/taxonomy.md

Approach and rejected alternatives

Modeled the document's shape directly on the two already-merged sibling standards (confidence.md, decision-references.md): Scope and authority / where the field's values come from / a MUST-numbered choosing procedure / enforcement table / exceptions and escalation / scope and omissions. Traced the enum's 13 values to PRD #602's own success-criteria bullet (confirmed by direct read, not by trusting the schema's paraphrase alone), rather than inventing a second, independently authored definition of each surface — the schema's own field description explicitly warns against that.

Rejected alternative: writing a one-line definition for each of the 13 enum values (architecture, layers, capabilities, etc.). No document in the repository defines these individually beyond the schema's shared one-line description, and manufacturing 13 new definitions would itself be the "second, independently invented taxonomy" the schema's description warns against. Instead the document gives a decision procedure grounded in how the enum has actually been used by the four nodes merged to origin/launchpad so far (three governance, one agent), and requires disclosing an imperfect fit in the node's own scope-and-omissions section — following the worked example already in corpus-readme.md.

Rejected relationship targets: the task brief suggested corpus-standard-status (#1323) as a plausible edge since both standards govern closed-enum fields. Checked git ls-tree -r --name-only origin/launchpad -- launchpad/docs/corpus directly rather than trusting the dispatch-time claim that #1323 and #1317 (corpus-standard-identifiers) were already merged — neither is. Both exist only on unmerged task branches (confirmed corpus-standard-status's commit is not an ancestor of origin/launchpad via git merge-base --is-ancestor). Per AGENTS.md step 9, a relationship target must exist on the branch being merged into, so neither is used. Declared one relationship instead: references → corpus-readme, justified because corpus-readme.md's own scope-and-omissions section is the concrete worked example this document cites (type: governance as "the closest true fit, not an exact one").

Verification

Command run:

python3 launchpad/project-intelligence/corpus/validate.py

Raw output:

UNVERIFIED  corpus-agents: evidence entry 1, citation 1: is a commit reference, which names no openable file
UNVERIFIED  corpus-agents: evidence entry 20, citation 1: is a graph-edge or tool-result citation, which names no openable file
UNVERIFIED  corpus-readme: evidence entry 1, citation 1: is a commit reference, which names no openable file
UNVERIFIED  corpus-standard-confidence: evidence entry 1, citation 1: is a commit reference, which names no openable file
UNVERIFIED  corpus-standard-decision-references: evidence entry 1, citation 1: is a commit reference, which names no openable file
UNVERIFIED  corpus-standard-taxonomy: evidence entry 1, citation 1: is a commit reference, which names no openable file
UNVERIFIED  corpus-standard-taxonomy: evidence entry 11, citation 1: is a graph-edge or tool-result citation, which names no openable file
PASS  corpus validation found no errors; 7 item(s) reported unverified

Exit status: 0

Also ran, as the local pre-commit verify-gate hook requires:

python3 -m unittest discover -s launchpad/docs/corpus/schema/tests -p "test_*.py" -v

Raw output (tail):

Ran 27 tests in 0.048s

OK

• Tests or checks were run and the raw output is pasted above
• The diff is confined to the scope of the linked issue
• No secrets, keys, tokens or hostnames were added to tracked files

Not verified

• Whether any generated view or index consumes/groups nodes by type — no generator exists yet in this corpus, so how a projection would render or rank an imperfect-fit disclosure is untested.
• Whether an author besides this one can actually apply the Choosing a value precedent-based heuristic in a case this document did not anticipate — the precedent sample is small (three governance nodes, one agent node).
• Whether corpus-standard-status (task: document corpus standard for status #1323) or corpus-standard-identifiers (task: document corpus standard for identifiers #1317), once actually merged to launchpad, will contain content that changes this document's reasoning — confirmed neither is merged at the checked revision, but their eventual content was not read for this document's purposes.
• No agent harness was tested reading this document as instructions; it is written and reviewed as corpus content only.

Security implications

None. This is a documentation-only change under launchpad/docs/corpus/standards/; it does not touch runtime code, secrets, or deployment configuration.

Escalations

None. The one open question flagged in the task brief — whether an edge to corpus-standard-status was warranted — resolved itself on verification: that node is not merged to origin/launchpad, so AGENTS.md's own rule rules it out without needing a judgment call.

[tucktuck101]

Review — standard for taxonomy (#1324)

The vocabulary is exactly right: the 13 type values match node.schema.json's closed enum
precisely — in order, nothing added, nothing missing, nothing renamed. For a taxonomy standard
that is the thing that had to be correct, and it is. The defect is structural.

Medium — no ## MUST section and no ## SHOULD section

The document's top-level headings, verbatim from grep -n '^## ':

Scope and authority / Where the 13 values come from / Choosing a value /
If none of the 13 fit / Enforcement, and where it stops /
Exceptions and escalation / Relationships / Scope and omissions

There is no ## MUST and no ## SHOULD. #1324's own definition-of-done carries the
MUST/SHOULD-split checkbox, and the meta-standard requires the two as separate sections — one
list with mixed modal verbs does not satisfy it however clearly the verbs are written. This is
the same remedy #1507 records against the sibling nodes and that #1508 is landing in this very
batch for confidence.md, so it is an established fix rather than a judgement call.

"Choosing a value" and "If none of the 13 fit" carry the normative content; splitting them into
## MUST / ## SHOULD preserves every identifier and cross-reference.

Batch correction — the AGENTS.md hunk is not an index registration

launchpad/docs/corpus/AGENTS.md at this head is byte-identical to the launchpad tip
(verified with git diff --quiet). The hunk is the task/636 change inherited through a stale
merge base (919886b41), not a registration this PR adds — merging produces no net AGENTS.md
change and there is no cross-standard conflict there. No standards index exists to register in.

What is correct

• 13-value type enum reproduced exactly from the schema — verified value by value.
• "If none of the 13 fit" routes to the schema's COMPATIBILITY.md change process rather than
  inventing a local escape, which is the right answer.
• The enforcement section states where the check stops instead of overstating it.
• CI green at head (latest run per check); node validates clean.

Reviewed by tucktuck101's review lane. Every failing claim reported above was reproduced by me against this PR head before posting.