mirrors/z3
3
0
Fork 0
mirror of https://github.com/Z3Prover/z3 synced 2026-06-10 10:57:15 +00:00
Code Activity

Add action/expectation/result structure to all skill definitions

Browse source 
Each step in every SKILL.md now carries labeled Action, Expectation,
and Result blocks so the agent can mechanically execute, verify, and
branch at each stage. Format chosen after comparing three variants
(indented blocks, inline keywords, tables) on a prove-validity
simulation; indented blocks scored highest on routing completeness
and checkability.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
This commit is contained in:
Angelica Moreira 2026-03-11 19:51:59 +00:00
parent d349b93d1d
commit 9d674404c8
10 changed files with 364 additions and 48 deletions
Download patch file Download diff file Expand all files Collapse all files

51
.github/skills/memory-safety/SKILL.md vendored
View file

@ -7,7 +7,22 @@ Build Z3 with compiler-based sanitizer instrumentation, execute the test suite,
# Step 1: Configure and build
The script invokes cmake with the appropriate `-fsanitize` flags and builds the `test-z3` target. Each sanitizer uses a separate build directory to avoid flag conflicts. If a prior instrumented build exists with matching flags, only incremental compilation runs.
Action:
Invoke the script with the desired sanitizer flag. The script calls
cmake with the appropriate `-fsanitize` flags and builds the `test-z3`
target. Each sanitizer uses a separate build directory to avoid flag
conflicts.
Expectation:
cmake configures successfully and make compiles the instrumented binary.
If a prior build exists with matching flags, only incremental compilation
runs.
Result:
On success: an instrumented `test-z3` binary is ready in the build
directory. Proceed to Step 2.
On failure: verify compiler support for the requested sanitizer flags
and review cmake output.
```bash
python3 scripts/memory_safety.py --sanitizer asan
@ -22,7 +37,21 @@ python3 scripts/memory_safety.py --sanitizer asan --skip-build --build-dir build
# Step 2: Run and collect
The test binary runs with `halt_on_error=0` so the sanitizer reports all violations rather than aborting on the first. The script parses `ERROR: AddressSanitizer`, `runtime error:`, and `ERROR: LeakSanitizer` patterns from the combined output, extracts source locations where available, and deduplicates by category, file, and line.
Action:
Execute the instrumented test binary with halt_on_error=0 so all
violations are reported rather than aborting on the first.
Expectation:
The script parses AddressSanitizer, UndefinedBehaviorSanitizer, and
LeakSanitizer patterns from combined output, extracts source locations,
and deduplicates by category/file/line.
Result:
On `clean`: no violations detected.
On `findings`: one or more violations found, each printed with severity,
category, message, and source location.
On `timeout`: test suite did not finish; increase timeout or investigate.
On `error`: build or execution failed before sanitizer output.
```bash
python3 scripts/memory_safety.py --sanitizer asan --timeout 900 --debug
@ -30,10 +59,20 @@ python3 scripts/memory_safety.py --sanitizer asan --timeout 900 --debug
# Step 3: Interpret results
- `clean`: no sanitizer violations detected.
- `findings`: one or more violations found. Each is printed with severity, category, message, and source location.
- `timeout`: the test suite did not complete within the deadline. Increase the timeout or investigate a possible infinite loop.
- `error`: build or execution failed before sanitizer output could be collected.
Action:
Review printed findings and query z3agent.db for historical comparison.
Expectation:
Each finding includes severity, category, message, and source location.
The database query returns prior runs for trend analysis.
Result:
On `clean`: no action required; proceed.
On `findings`: triage by severity and category. Compare against prior
runs to distinguish new regressions from known issues.
On `timeout`: increase the deadline or investigate a possible infinite
loop.
On `error`: inspect build logs before re-running.
Query past runs:
```bash