[DanielNetzer]

Implements issue #299.

Molecule: mol-299-59f9d54feba2fc546f3001f0
Formula: forgecell-github-issue
Base: 8b0a3cb561abafe129e04da177a056cde4a9b39f
Delivery: efbcd6dcd76b595170483110bbff6bbd3d55e905

Awaiting exact-commit CI and human review. Harness-reported checks are not independent CI evidence. No merge, deployment or issue closure is authorized.

[DanielNetzer]

Adversarial review (plugins slice 2b-3): no HIGH findings and no GitHub-path drift. I checked:

• mol-%d-, issue-%d- and the numeric lock format;
• that plan JSON and digests for numeric Issues are unchanged (Key is omitempty everywhere);
• that "Implements issue #9." is byte-identical;
• that keyed identity is the raw Key, never the slug, so ENG-1 and eng_1 can't merge in queue, history or locks;
• that the slug can't form a path;
• that RunKeyed refuses Approve.

Two MEDIUMs were fixed in this PR, with one verify pass:

1. freshIntakeAction keeps its Number >= 1 guard. A keyed blocked Molecule would otherwise show a "Start a fresh intake" action that posts issue: 0.
2. ParseIssueKey compiles the raw pattern before wrapping it, so ^x)|(y$ is refused rather than turned into a top-level alternation.

Deferred to slice 2b-4, where keyed runs first become reachable:

• the keyed renderers cli.go:1179 (forgecell run 0 … --approve), cli.go:1890 (repo#0) and ui/projection.go:588 (#0);
• a keyed freshIntakeAction;
• wiring issueKeyLockName into any keyed loop.

LOWs:

• readiness/plan.go:43: key validation (512 bytes, IsControl only) is looser than intake.ValidateIssueKey, and doesn't require a usable slug.
• Purely numeric keys: a key such as "12" gives mol-12-<nonce>, which looks like a numeric Molecule. It doesn't collide, because of the nonce, and nothing parses ids. Consider a mol-k- prefix.

Integration: I applied this branch to current main (with #300). go vet ./... passes, and every package compiles its tests.