Skip to content

Commit f7063e4

Browse files
docs(ahbg): extract calibration evidence receipts from closed #18/#19
Carry the comparison results, corpus receipts, and reproducible calibration evidence needed by canonical AHBG. Provider workspace trees under stack/ahbg/codex and stack/ahbg/deepseek are not merged; closed branches are treated as experiment receipts only. Records frozen build SHAs (Grok cce9cec, Codex ffb64c2, DeepCode ec07f46), the 35/35 successor runs, all six directional SURVIVED checks, the successor corpus receipt (1.0.1-proposal-1, scenario digest 371d2361), the canonical main corpus path/digest (bc521113ffa7), and preserved evidence caveats.
1 parent 7c4f1c1 commit f7063e4

2 files changed

Lines changed: 205 additions & 0 deletions

File tree

Lines changed: 131 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,131 @@
1+
{
2+
"schema": "interdependency.ahbg.calibration-evidence/1.0.0",
3+
"standing": "experiment-receipts-not-production-workspaces",
4+
"extracted_from": {
5+
"closed_prs": [18, 19],
6+
"branch_heads_at_extraction": {
7+
"agent/ahbg-codex": "af67642f987124f91aad63e4583422602ed508eb",
8+
"agent/ahbg-deepcode": "34648ed63e8e41c8dc775ecc59a253257c5a9de8",
9+
"agent/ahbg-grok": "03e77a33f2a79bc586c25b17e04147a29a187218"
10+
},
11+
"policy": "provider workspace trees under stack/ahbg/codex and stack/ahbg/deepseek are not merged into main; only comparison results, corpus receipts, and reproducible calibration evidence are carried here."
12+
},
13+
"builders": [
14+
{
15+
"builder": "Grok",
16+
"branch": "agent/ahbg-grok",
17+
"frozen_build_sha": "cce9cec7dae61304118efcd47bc0d7461200d335",
18+
"successor_corpus_run": {
19+
"scenarios": 35,
20+
"survived": 35,
21+
"falsified": 0,
22+
"unresolved": 0,
23+
"blocked": 0
24+
}
25+
},
26+
{
27+
"builder": "Codex",
28+
"branch": "agent/ahbg-codex",
29+
"frozen_build_sha": "ffb64c274583d8539f8f4fe7e0aa77366689e910",
30+
"freeze_verification": {
31+
"a0_tests": "7 OK",
32+
"ahbg_tests": "12 OK",
33+
"python_compile": "15 files OK",
34+
"artifact_check": "SURVIVED",
35+
"calibration_smoke_summary": {
36+
"survived": 4,
37+
"falsified": 0,
38+
"unresolved": 2,
39+
"blocked": 0
40+
}
41+
},
42+
"successor_corpus_run": {
43+
"scenarios": 35,
44+
"survived": 35,
45+
"falsified": 0,
46+
"unresolved": 0,
47+
"blocked": 0
48+
},
49+
"evidence_caveat": "Codex recorded that its conversation inspected sibling DeepCode/DeepSeek files before the workspace correction; the frozen implementation lives under stack/ahbg/codex and does not import sibling code."
50+
},
51+
{
52+
"builder": "DeepCode",
53+
"branch": "agent/ahbg-deepcode",
54+
"frozen_build_sha": "ec07f465184e7a37af856bc5b301bd8eaa4f097b",
55+
"coordination_base_commit": "8fd2923292361b1956a003bd5c74eae50a5323b0",
56+
"successor_corpus_run": {
57+
"scenarios": 35,
58+
"survived": 35,
59+
"falsified": 0,
60+
"unresolved": 0,
61+
"blocked": 0
62+
}
63+
}
64+
],
65+
"corpus_receipt": {
66+
"corpus_id": "calibration-family",
67+
"proposal_version": "1.0.1-proposal-1",
68+
"scenario_count": 35,
69+
"canonical_scenarios_sha256": "371d2361f57b56d73544f58b247704617d550a7a0685a133c4f8b1ff3b36c835",
70+
"corpus_json_sha256_at_successor_runs": "ea172cb68a1a31be843f45c9886590f95f60daad4f10b9e42732bfd416ef73ab",
71+
"canonical_main_corpus_file_sha256": "bc521113ffa7bd6d5094c71f3ad66547d5f00260f380258e43c2086533a5d7ed",
72+
"canonical_main_corpus_path": "ahbg/deepseek/corpus-proposal/corpus.json",
73+
"note": "canonical main re-serialized the corpus file after the successor runs; the scenario digest is unchanged. Both file digests are recorded so the successor-run receipt remains reproducible.",
74+
"predecessor": {
75+
"proposal_version": "1.0.0-proposal-1",
76+
"canonical_scenarios_sha256": "b05cba2cf2f15583548cc15158f09e2612545c978b6a42ddeb314f1e4ed0e5e0",
77+
"merged_to_main_as": "3a92c7b0f8568e6fc2600b45bca760030ea2ba3f",
78+
"merged_via_pr": 5
79+
}
80+
},
81+
"directional_checks": [
82+
{
83+
"checker": "Grok",
84+
"subject": "Codex",
85+
"subject_frozen_build": "ffb64c274583d8539f8f4fe7e0aa77366689e910",
86+
"standing": "SURVIVED",
87+
"checker_artifact": "ahbg/grok/reviews/codex-review.json"
88+
},
89+
{
90+
"checker": "Grok",
91+
"subject": "DeepCode",
92+
"subject_frozen_build": "ec07f465184e7a37af856bc5b301bd8eaa4f097b",
93+
"standing": "SURVIVED",
94+
"checker_artifact": "ahbg/grok/reviews/deepcode-review.json"
95+
},
96+
{
97+
"checker": "Codex",
98+
"subject": "Grok",
99+
"subject_frozen_build": "cce9cec7dae61304118efcd47bc0d7461200d335",
100+
"standing": "SURVIVED",
101+
"checker_artifact": "ahbg/codex/reviews/grok-review.json"
102+
},
103+
{
104+
"checker": "Codex",
105+
"subject": "DeepCode",
106+
"subject_frozen_build": "ec07f465184e7a37af856bc5b301bd8eaa4f097b",
107+
"standing": "SURVIVED",
108+
"checker_artifact": "ahbg/codex/reviews/deepcode-review.json"
109+
},
110+
{
111+
"checker": "DeepCode",
112+
"subject": "Grok",
113+
"subject_frozen_build": "cce9cec7dae61304118efcd47bc0d7461200d335",
114+
"standing": "SURVIVED",
115+
"checker_artifact": "ahbg/deepseek/reviews/grok-review.json"
116+
},
117+
{
118+
"checker": "DeepCode",
119+
"subject": "Codex",
120+
"subject_frozen_build": "ffb64c274583d8539f8f4fe7e0aa77366689e910",
121+
"standing": "SURVIVED",
122+
"checker_artifact": "ahbg/deepseek/reviews/codex-review.json"
123+
}
124+
],
125+
"hmmm": [
126+
"world-digest mismatches remain 35 of 35 in the burden-coupling comparison (expected while standing is binding);",
127+
"formal sealing of the successor corpus is recorded as adopted on main, not as a three-way branch merge;",
128+
"Codex evidence caveat about inspecting sibling files before workspace correction is preserved and not erased;",
129+
"provider workspace trees on the closed branches are receipts only and are intentionally not merged."
130+
]
131+
}
Lines changed: 74 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,74 @@
1+
# AHBG calibration evidence
2+
3+
Extracted from the closed calibration PRs #18 (Codex) and #19 (DeepCode) on
4+
2026-09-01. The provider workspace trees (`stack/ahbg/codex`,
5+
`stack/ahbg/deepseek`) remain branch-resident experiment receipts; only the
6+
comparison results, corpus receipts, and reproducible calibration evidence are
7+
carried here.
8+
9+
Machine-readable record: [`CALIBRATION_EVIDENCE.json`](CALIBRATION_EVIDENCE.json).
10+
11+
## Frozen builds
12+
13+
| Builder | Branch | Frozen build SHA |
14+
|---|---|---|
15+
| Grok | `agent/ahbg-grok` | `cce9cec7dae61304118efcd47bc0d7461200d335` |
16+
| Codex | `agent/ahbg-codex` | `ffb64c274583d8539f8f4fe7e0aa77366689e910` |
17+
| DeepCode | `agent/ahbg-deepcode` | `ec07f465184e7a37af856bc5b301bd8eaa4f097b` |
18+
19+
Branch heads at extraction: Codex `af67642`, DeepCode `34648ed`, Grok `03e77a3`.
20+
21+
## Successor corpus receipt
22+
23+
- Corpus `calibration-family`, proposal `1.0.1-proposal-1`, 35 scenarios.
24+
- Canonical scenarios digest:
25+
`371d2361f57b56d73544f58b247704617d550a7a0685a133c4f8b1ff3b36c835`.
26+
- Corpus JSON file digest at the successor runs:
27+
`ea172cb68a1a31be843f45c9886590f95f60daad4f10b9e42732bfd416ef73ab`.
28+
- Canonical main re-serialized the corpus file; current
29+
`ahbg/deepseek/corpus-proposal/corpus.json` file digest is
30+
`bc521113ffa7bd6d5094c71f3ad66547d5f00260f380258e43c2086533a5d7ed`
31+
with the same scenario digest.
32+
- Predecessor `1.0.0-proposal-1`, digest
33+
`b05cba2cf2f15583548cc15158f09e2612545c978b6a42ddeb314f1e4ed0e5e0`,
34+
merged to main as `3a92c7b0` via PR #5.
35+
36+
## Successor runs
37+
38+
All three builders: **35 SURVIVED / 0 FALSIFIED / 0 UNRESOLVED / 0 BLOCKED**.
39+
40+
Codex freeze verification: `a0` tests 7 OK, `ahbg` tests 12 OK, Python
41+
compile 15 files OK, artifact check SURVIVED, calibration smoke summary
42+
4 survived / 0 falsified / 2 unresolved / 0 blocked.
43+
44+
## Six directional checks — all SURVIVED
45+
46+
| Checker | Subject | Subject frozen build | Checker artifact |
47+
|---|---|---|---|
48+
| Grok | Codex | `ffb64c2` | `ahbg/grok/reviews/codex-review.json` |
49+
| Grok | DeepCode | `ec07f46` | `ahbg/grok/reviews/deepcode-review.json` |
50+
| Codex | Grok | `cce9cec` | `ahbg/codex/reviews/grok-review.json` |
51+
| Codex | DeepCode | `ec07f46` | `ahbg/codex/reviews/deepcode-review.json` |
52+
| DeepCode | Grok | `cce9cec` | `ahbg/deepseek/reviews/grok-review.json` |
53+
| DeepCode | Codex | `ffb64c2` | `ahbg/deepseek/reviews/codex-review.json` |
54+
55+
No builder was falsified. Agreement across independent builds is replication
56+
evidence, not proof.
57+
58+
## Evidence caveats (preserved, not erased)
59+
60+
- Codex recorded that its conversation inspected sibling DeepCode/DeepSeek
61+
files before the workspace correction; the frozen implementation lives under
62+
`stack/ahbg/codex` and does not import sibling code.
63+
- DeepCode's frozen build was reviewed by Grok and Codex; later commits on the
64+
branch are metadata, extension evidence, and reciprocal-review records that
65+
do not alter the frozen implementation.
66+
67+
## hmmm
68+
69+
- World-digest mismatches remain 35 of 35 in the burden-coupling comparison
70+
(expected while standing is binding).
71+
- Formal successor sealing is recorded as adopted on main, not as a three-way
72+
branch merge.
73+
- Provider workspace trees on the closed branches are receipts only and are
74+
intentionally not merged into main.

0 commit comments

Comments
 (0)