This file is a dated, signed snapshot of what can honestly be claimed about trigger at one exact commit. It is the third file of the README / EXPLAINME / AFFIRMATION trio. The README says where the project is going. EXPLAINME says how it is built. This file says what was true and checkable at the anchor below.
Every claim here comes from running the project’s own checks in the session that wrote this file. Where the README, the Justfile or memory disagree with a live run, the live run wins and the disagreement is recorded.
The owner lands this file with a signed git commit. That signature over this content, at the anchored commit, is what makes the affirmation attributable.
This document records best belief at the timestamp below. It is not a guarantee of correctness. The only guarantee is no intentional overclaim. Where something was checked, it says how. Where something was not checked, it says so. A claim that later proves false is an error to fix, not a lie.
You are invited to refute any claim here with a failing run, a counter-example or a contradicting source.
| Field | Value |
|---|---|
Repo |
|
Branch |
|
Commit (HEAD) |
|
Permalink |
https://github.com/hyperpolymath/trigger/tree/0755f3b99b0ee103b554aa1c3d5fd515dcca301b |
Verified (UTC) |
2026-10-07T12:02:03Z |
Working-tree delta at verification |
none ( |
Toolchain |
GNAT 14.2.0 + gprbuild (Debian); Zig 0.16.0; Idris 2 0.7.0; Agda 2.7.0.1 with agda-stdlib 2ffa8b7d; just 1.56.0. Not available: gnatprove, alr. |
If you are reading this at a later commit, the claims may have drifted. Re-run the reproduction steps and write a fresh affirmation; do not trust a stale one.
-
README.adocdescribes an Ada/SPARK implementation with Zig FFI bindings and an Idris2 API layer. The runs below show that none of those three layers builds at the anchor with the toolchain listed. -
The GitHub repository description, as of the anchor, calls trigger a social media reporting utility with multi-account management, persistent sessions, proxy support, FloodWait handling and a TUI. That is a feature statement. Nothing below verifies any of those features.
-
This file replaces an earlier
AFFIRMATION.adocthat was the unfilled estate template (placeholder heading and anchor fields). It made no claims, so it is replaced in place rather than archived. Git history keeps it.
At the anchor, trigger does not build. The Ada project file is rejected, the Justfile does not parse, every Zig FFI module fails to compile, every Idris2 module fails to resolve an import, and both Agda proof modules are rejected. Nothing about runtime behaviour was or could be verified.
-
The repository is clean at the anchor and its history is intact.
-
Six of the ten check contexts on the anchor commit concluded
SUCCESS(read from the commit’s status check rollup over the GitHub GraphQL API): Hypatia neurosymbolic analysis, panic-attack assail, the rust-secrets and shell-secrets scans, Patch Bridge CVE triage, and the gitbot-fleet deposit. These are scanners over the tree, not evidence that the code works.
-
"Fails here" is not "never worked". Each failure below is measured at the toolchain versions in the anchor. Two of them look like code written for an older toolchain:
-
The Zig
telegrammodule calls@ptrCastand@alignCastwith two arguments. Current Zig takes one. -
Every Idris2 module imports
Control.Monad.Effect, which is the Idris 1 Effects library and does not ship with Idris 2.
-
-
Some failures do not depend on the version:
-
The Agda modules declare
--safeand also contain postulates (8 inTrigger/API.agda, 5 inTrigger/Epistemic.agda). Agda rejects that combination in every version. -
Four Zig modules fail at the parser, for example
@null,incrypto/crypto.zig:121.
-
-
SPARK proofs were not checked, rather than found failing.
gnatproveis not installed here, and the project file does not process, so the SPARK layer is untested at this anchor. -
The Ada tests were not runnable. The test target goes through the same project file that fails to process.
-
The two Hypatia contexts on the anchor disagree: "Hypatia / Hypatia Neurosymbolic Analysis" concluded
SUCCESSand "Hypatia neurosymbolic scan" concludedFAILURE. Their logs were not read for this file. -
scan / gitleaksconcludedFAILURE. The job’s log was not read, so this file makes no claim about whether it found anything.
| Check | Result at the anchor |
|---|---|
|
Refuted. |
|
Refuted. rc=5, |
Ada/SPARK CI workflow on the anchor commit |
Concluded |
Zig FFI, |
Refuted, 5 of 5. Parse errors in |
Idris2 API, |
Refuted, 5 of 5: |
Agda, |
Refuted, 2 of 2, rc=42: |
Other CI on the anchor |
Rollup contexts at |
From a clone checked out at the anchor commit:
git checkout 0755f3b99b0ee103b554aa1c3d5fd515dcca301b
just build # Justfile parse error at 223
gprbuild -P trigger.gpr -XLIBRARY_TYPE=static # rc=5, project file rejected
for d in crypto discord social_media twitter telegram; do
(cd ffi/zig/$d && for f in *.zig; do zig build-lib -fno-emit-bin "$f"; done)
done
(cd ffi/idris2 && for f in *.idr; do idris2 --check "$f"; done) # look for "Error:"
(cd proofs/agda && agda Trigger/API.agda && agda Trigger/Epistemic.agda)Agda needs agda-stdlib. Point AGDA_DIR at a directory whose libraries file
names the stdlib’s .agda-lib and whose defaults file says
standard-library.
At 0755f3b, trigger is an unbuilt Ada/SPARK, Zig and Idris2 codebase: every
language layer and both Agda proofs fail their own checks, so no functional
claim is affirmed.
Engineering party (AI). Claude (claude-opus-5-5), session run at
2026-10-07T12:02:03Z, ran every check above and wrote this file. The wording
is a faithful report of those runs. Nothing was copied forward from an earlier
affirmation.
Owner. Jonathan D.A. Jewell signs by committing this file with an SSH signing key:
git commit -S -s AFFIRMATION.adoc -m "docs: affirm state at 0755f3b9"
git log --show-signature -1