Remove ASTRA narrative field from skills; bump astra-tools to 0.2.10

[EiffL]

ASTRA's narrative field has been removed from the spec (astra-spec RFC-0002) — replaced by a single optional description on any Analysis, with rich write-ups now authored as external reports that reference analysis elements by their tree-path. This brings the skill bundle in line.

Dependency

• astra-tools>=0.2.5 → >=0.2.10 (the release that requires astra-spec>=0.0.11, the narrative-free schema).

Skills

• Deleted the narrative skill (SKILL.md + references) and every pointer to it: docs mirror (docs/skills/narrative.md), zensical.toml nav, README/index rows, CLAUDE.md bundle note. The paper-reproduction bundle is now five skills (four siblings).
• astra (spec reference): dropped narrative/authors from the field list, removed narrative from the reserved-ID set, and replaced the ## Narrative section with a ## Description + element-addressing section.
• lc-new: "Populate Narrative" → "Populate Description".
• lc-from-paper (SKILL + architect/specify/implement references): ARCHITECT authors a short per-analysis description (no five-section narrative, no astra-anchor: weaving); SPECIFY drops the /narrative substrate invocation and authors rationale:/claim:/notes: directly on entries.
• paper-extraction: the extract-paper-substrate.py script now emits a top-level description: (from the abstract) instead of narrative.summary, and no longer emits a narrative.findings cross-link; the SKILL prose and the example YAML are aligned (example bumped to 0.0.11).
• figure-comparison, lc-from-code, check-sentence-by-sentence: source description/findings: per-element fields instead of narrative.*.
• docs/user: FINALIZE authors a description:.

Remaining uses of the word "narrative" are general-English (constitution/report writing, console output) — not the ASTRA field.

Verification

• uv run pytest → 335 passed
• The paper-extraction example and the script's generated stub both validate against astra-spec 0.0.11
• No new ruff findings (CI lints src/ tests/; the skill script is out of scope); no dangling links to the removed skill

🤖 Generated with Claude Code

[github-actions]

✅ Eval Results

Metric	Value
Score	1.00
Build complete	✅
Cost	$0.83
Turns	30
Duration	381s
lightcone-cli	0.3.7.dev8+g3f3255db3 (3f3255db)
Results	Download

Graders

✅ spec_valid (1.00)
✅ all_materialized (1.00)

Full output

ystow-0.8.19 pytest-9.1.1 pytest-logging-2015.11.4 pyyaml-6.0.3 rapidfuzz-3.14.5 rdflib-7.6.0 referencing-0.37.0 requests-2.34.2 rich-15.0.0 rpds-py-2026.6.3 smart-open-7.7.1 smmap-5.0.3 snakemake-9.23.1 snakemake-interface-common-1.23.0 snakemake-interface-executor-plugins-9.4.0 snakemake-interface-logger-plugins-2.1.0 snakemake-interface-report-plugins-1.3.0 snakemake-interface-scheduler-plugins-2.0.2 snakemake-interface-storage-plugins-4.4.1 sortedcontainers-2.4.0 sqlmodel-0.0.37 tabulate-0.10.0 tblib-3.2.2 tenacity-9.1.4 throttler-1.2.3 toolz-1.1.0 tornado-6.5.7 tqdm-4.68.3 traitlets-5.15.1 typing-extensions-4.15.0 typing-inspection-0.4.2 urllib3-2.7.0 wrapt-2.2.2 yte-1.9.4 zict-3.0.0
08:53:00 lightcone.eval.sandbox [sandbox build] WARNING: Running pip as the 'root' user can result in broken permissions and conflicting behaviour with the system package manager, possibly rendering your system unusable. It is recommended to use a virtual environment instead: https://pip.pypa.io/warnings/venv. Use the --root-user-action option if you know what you are doing and want to suppress this warning.
08:53:01 lightcone.eval.sandbox [sandbox build] #13 [10/11] WORKDIR /home/evaluser
08:53:01 lightcone.eval.sandbox [sandbox build] #14 [11/11] RUN mkdir -p ~/.claude && echo '{"hasCompletedOnboarding": true}' > ~/.claude.json
08:53:01 lightcone.eval.sandbox [sandbox build] #15 exporting to image
08:53:23 lightcone.eval.sandbox Created sandbox ed902bbb-2640-42a5-aff0-011280b5e198 for trial build-snae-0
08:53:23 httpx HTTP Request: POST https://proxy.app.daytona.io/toolbox/ed902bbb-2640-42a5-aff0-011280b5e198/files/bulk-upload "HTTP/1.1 200 OK"
08:53:24 lightcone.eval.sandbox Installed wheels: ['lightcone_cli-0.3.7.dev8+g3f3255db3-py3-none-any.whl']
08:53:25 httpx HTTP Request: POST https://proxy.app.daytona.io/toolbox/ed902bbb-2640-42a5-aff0-011280b5e198/files/bulk-upload "HTTP/1.1 200 OK"
08:53:25 httpx HTTP Request: POST https://proxy.app.daytona.io/toolbox/ed902bbb-2640-42a5-aff0-011280b5e198/files/bulk-upload "HTTP/1.1 200 OK"
08:53:25 httpx HTTP Request: POST https://proxy.app.daytona.io/toolbox/ed902bbb-2640-42a5-aff0-011280b5e198/files/bulk-upload "HTTP/1.1 200 OK"
08:53:25 httpx HTTP Request: POST https://proxy.app.daytona.io/toolbox/ed902bbb-2640-42a5-aff0-011280b5e198/files/bulk-upload "HTTP/1.1 200 OK"
08:59:48 lightcone.eval.sandbox Deleted sandbox for trial build-snae-0
  snae trial 0: score=1.00 complete

lightcone-cli: 0.3.7.dev8+g3f3255db3 (HEAD 3f3255db)
ASTRA: 0.2.10

  Eval Results: Scores  
┏━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Task ┃     Score     ┃
┡━━━━━━╇━━━━━━━━━━━━━━━┩
│ snae │ 1.00 +/- 0.00 │
│      │ pass@k: 100%  │
└──────┴───────────────┘

   Eval Results: Cost &   
         Duration         
┏━━━━━━┳━━━━━━━━━━━━━━━━━┓
┃ Task ┃ Cost / Duration ┃
┡━━━━━━╇━━━━━━━━━━━━━━━━━┩
│ snae │      $0.83      │
│      │      381s       │
└──────┴─────────────────┘

Total: 1 trials, $0.83, 381s

Results saved to: eval-results/build-3f3255db/results.json

[cloudflare-workers-and-pages]

Deploying with    Cloudflare Workers

The latest updates on your project. Learn more about integrating Git with Workers.

Status	Name	Latest Commit	Preview URL	Updated (UTC)
✅ Deployment successful! View logs	lightcone-cli	ea3c6c0	Commit Preview URL Branch Preview URL	Jun 30 2026, 08:34 AM

[EiffL]

a few suggestions

[cailmdaley]

hi François this looks great. as you raised on Slack i think we should replace the narrative skill with a different one to author these new MyST documents, and i would suggest doing it in this PR since a lot of the old narrative skill instructions will be applicable to MyST documents and can be transferred over to the new skill.

[cailmdaley]

oops, this file, the examples folder, and references to it should be deleted entirely. i should have caught it when i added this skill, it's an anecdote that i don't think will improve agent performance.

[EiffL]

hum ok, thanks, removing it

[EiffL]

Ok thanks Cail, I thought about it a bit more, mystra is not fully mature yet, so I think it's better to remove the narrative skill for now, and add a nice mystra skill once the first stable release of mystra comes out. In the meantime we are in any case not providing a system for the users to visualize the narrative fields, so from an end-user perspective nothing has changed.

[EiffL]

LGTM