[d6veteran]

F1 from the releases page, approved as D1 on 2026-11-05. The points come from the effort evidence.

What changes

• Job words in Who. A row reads <Persona>, <job>. Five job words each name a model and an effort, one job line apiece in tiers.conf:
  • design: Opus 5.5 at high
  • build, review, write: Opus 5.5 at medium
  • edit: Sonnet 5.5 at medium
• Pinning. <Persona>, <job>, <Model> <effort> pins any other point, for example Akira, design, Fable high.
• Older plans keep working. <Persona>, <Model> is read by its tier: Fable as design, Opus as build, Sonnet or Haiku as edit. check adds one note line saying how many rows use that form. No plan file is rewritten.
• Cards and the board. plan.py card names the session [P2-S4] [Akira design] <title> and prints job, the model's id and effort. The card's model check now covers effort too. A build card carries the line "As much coded as necessary, as little code as possible." rows and ready print the point, such as Opus high.
• Opening a stage. /ck:start passes model and effort on the chip. A chip can't open above the session that opens it on either axis. When a stage's point is above that session's, the session says so and the stage's own check stops it. A session can read its own effort with get_session and self; I checked this before building. Refreshing the board now sets a running stage's effort as well as its model.
• The Fable rule. The plan contract and the draft workflow pin Fable high only for a stage that Opus at high has already fallen short on, or, on trial, for security or concurrency code. They never pin xhigh or max unless asked.
• How we work 0.1.27. The session tag is [{persona} {job}]. "The stage's model and effort" lists the five points in words, for briefs that name none.
• Cost by point. cost.py keeps each reply's effort and logs cost by model and effort with each pull request. /ck:cost history adds a cost-per-point table for pull requests merged from 0.1.27 on.

Tests

• tests/run.sh: 504 passed, 0 failed.
• tests/plan-cases.py: 194 ok, 0 failed. This includes new cases for each job word, a pinned point, an unknown job word and the older-form note.
• The plan fixtures now use job words.
• cost.py history was checked against a hand-made log: the cost per point is each session's change between rows.

Not in this release

• d20mob and hireling still have their own How we work copies, which read the tag's second word as a model. They pick up 0.1.27 through /ck:claude-md on your word; until then, the check on ck's own cards covers them.
• Not done yet: re-running a failed build at high, moving the six Fable personas in tiers.conf, and job words in /ck:do.

🤖 Generated with Claude Code

[d6veteran]

ck cost for this pull request: $12.51, built by 0.1.27 groundwork: evidence on models and effort.

0.1.27 groundwork: evidence on models and effort cost summary

Lane	Model	Input	Output	Cache write	Cache read	Total
main	opus-5-5	324 · $0.00	120k · $2.39	495k · $3.96	30.2M · $6.03	30.8M · $12.39
web-fetch	haiku-5-5	34 · $0.00	44k · $0.05	252k · $0.06	512k · $0.02	807k · $0.12
All		358 · $0.00	163k · $2.44	747k · $4.02	30.7M · $6.05	31.6M · $12.51

Prices from https://platform.claude.com/docs/en/about-claude/pricing.md, as of 2026-11-05.
Claude Code wrote no cost record for this session, so background calls are not counted.