Following #64, I measured the published ASu-skills v0.4.0 bundle against the codex_cli target with context-report v0.1.0.
Result
- Subject SHA-256:
5022d06b3b9fc2a39e30de575e0edeb3bb55549564f9bb92780335538b2bb2d5
- Client version:
26.901.51231
- The generated in-toto statement passed the producer's well-formed/bound verification.
cost.context_tokens: PASSED — 8,155 approximate tokens across 9 SKILL.md files using approx-regex-v1.conformance, decision, interference, and efficacy: NotAvailable.- Hook reachability, hook fault cases, and hook latency: NotApplicable, because this plugin declares no hooks.
Compared with the ASu-skills v0.3.0 / Claude Code sample in #68, this bundle contains 3 more measured skill files and 3,264 more approximate tokens (+66.7%). This is a static count of the complete skill bodies, not a claim that Codex loads all 8,155 tokens into every request; Codex skills are loaded on demand.
Reproduction
context-report produce <asu-skills-v0.4.0> --target codex_cli --client-version 26.901.51231 --n 20 --out asu-skills-v0.4.0-codex-cli-context-report.json
context-report verify asu-skills-v0.4.0-codex-cli-context-report.json
Measured files
| Skill file |
Approx. tokens |
| contributor |
1,253 |
| evidence-recap |
526 |
| great-resume |
408 |
| interview |
911 |
| job-apply |
970 |
| job-match |
499 |
| make-resume |
1,547 |
| offer |
356 |
| project-guide |
1,685 |
| Total |
8,155 |
Attestation binding
- Predicate:
https://open-coder-ai.github.io/context-report/attestation/v0.1
- Producer:
https://github.com/open-coder-ai/context-report#local
- Producer version:
context-report 0.1.0
- Produce-args SHA-256:
2e0428feaf7529746b02dbf9d1cc3ae3cec2881e7bec2a7eb0eebfd9f9361bed
- Context-token input SHA-256:
3cdc6319ba3809680c904990bbd6d4d942eeb16a2d726ac6583038600ce2e9e7
- Recorded at:
2026-09-09T11:39:15Z
Disclosure
AI assistance was used to run the documented measurement workflow and draft this issue. I reviewed the generated statement and the text above.
Following #64, I measured the published ASu-skills v0.4.0 bundle against the
codex_clitarget with context-report v0.1.0.Result
5022d06b3b9fc2a39e30de575e0edeb3bb55549564f9bb92780335538b2bb2d526.901.51231cost.context_tokens: PASSED — 8,155 approximate tokens across 9SKILL.mdfiles usingapprox-regex-v1.conformance,decision,interference, andefficacy: NotAvailable.Compared with the ASu-skills v0.3.0 / Claude Code sample in #68, this bundle contains 3 more measured skill files and 3,264 more approximate tokens (+66.7%). This is a static count of the complete skill bodies, not a claim that Codex loads all 8,155 tokens into every request; Codex skills are loaded on demand.
Reproduction
Measured files
Attestation binding
https://open-coder-ai.github.io/context-report/attestation/v0.1https://github.com/open-coder-ai/context-report#localcontext-report 0.1.02e0428feaf7529746b02dbf9d1cc3ae3cec2881e7bec2a7eb0eebfd9f9361bed3cdc6319ba3809680c904990bbd6d4d942eeb16a2d726ac6583038600ce2e9e72026-09-09T11:39:15ZDisclosure
AI assistance was used to run the documented measurement workflow and draft this issue. I reviewed the generated statement and the text above.