agent-loop-ext
Crash-resilient external agent loop with state persistence and CI/CD integration
Set or report the reproducibility mode governing how AIWG workflows execute, from standard to fully auditable
$ npx -y skills add jmagly/aiwg --skill execution-mode --agent claude-codeHow it fires
How this skill gets triggered: by you, by Claude, or both.
/execution-modeContext preview
The summary Claude sees to decide when to auto-load this skill.
Set or report the reproducibility mode governing how AIWG workflows execute, from standard to fully auditable
namespace: aiwg name: execution-mode platforms: [all] description: Set or report the reproducibility mode governing how AIWG workflows execute, from standard to fully auditable
You set or report the reproducibility mode that governs how AIWG workflows execute, controlling determinism constraints from none (standard) to fully auditable (audit).
Alternate expressions and non-obvious activations (primary phrases are matched automatically from the skill description):
| Pattern | Example | Action | |---------|---------|--------| | Set standard | "use default execution mode" | Run `aiwg execution-mode standard` | | Set seeded | "use a fixed seed for this run" | Run `aiwg execution-mode seeded --seed <value>` | | Set strict | "make execution deterministic" | Run `aiwg execution-mode strict --seed <value>` | | Set audit | "enable audit mode" | Run `aiwg execution-mode audit --seed <value>` | | Status check | "what execution mode is active?" | Run `aiwg execution-mode` (no args) | | Seed only | "set seed to 42" | Run `aiwg execution-mode seeded --seed 42` |
When triggered:
1. **Extract intent**:
2. **Run the appropriate command**:
# Report current mode (no args) aiwg execution-mode # Standard — no reproducibility constraints aiwg execution-mode standard # Seeded — fixed random seed for all operations aiwg execution-mode seeded --seed 42 # Strict — seeded + no external calls + pinned versions aiwg execution-mode strict --seed 42 # Audit — strict + decision logging for replay aiwg execution-mode audit --seed 42
3. **Mode summary**:
| Mode | Seed | External calls | Pinned versions | Decision log | |------|------|----------------|-----------------|--------------| | `standard` | No | Allowed | No | No | | `seeded` | Yes | Allowed | No | No | | `strict` | Yes | Blocked | Yes | No | | `audit` | Yes | Blocked | Yes | Yes |
4. **Persistence**: Mode is written to `.aiwg/execution-mode.json` and persists across sessions until explicitly changed.
5. **Report the result** — confirm the mode that is now active and note any implications (e.g., external calls will be blocked in strict/audit mode).
**User**: "What execution mode are we in?"
**Extraction**: Status check, no change requested
**Action**:
aiwg execution-mode
**Response**: "Current execution mode: `seeded` (seed: 1234). External calls are allowed and versions are not pinned. To increase determinism, switch to `strict` or `audit` mode."
**User**: "Make the workflow fully deterministic with seed 42"
**Extraction**: Strict mode requested, seed 42 provided
**Action**:
aiwg execution-mode strict --seed 42
**Response**: "Execution mode set to `strict` with seed 42. External network calls are blocked and dependency versions are pinned. Mode saved to `.aiwg/execution-mode.json`."
**User**: "Enable audit mode so we can replay every decision"
**Extraction**: Audit mode requested, no seed provided — generate one
**Action**:
aiwg execution-mode audit --seed 7391
**Response**: "Execution mode set to `audit` with seed 7391. All decisions will be logged for replay. Seed chosen automatically — note it if you need to reproduce this run exactly."
**User**: "Reset execution mode back to default"
**Extraction**: Reset to standard mode
**Action**:
aiwg execution-mode standard
**Response**: "Execution mode reset to `standard`. No reproducibility constraints are active."
If the user's intent is ambiguous:
Reusable project context and specialist workflows for the AI tools you already use. Plan software, coordinate specialist reviews, prepare campaigns, investigate incidents, organize research, curate media, and maintain operational knowledge.
Repo: jmagly/aiwg
Crash-resilient external agent loop with state persistence and CI/CD integration
Detect requests for iterative autonomous agent loops and route to the appropriate loop executor
Automatically execute tests when code-generating agents modify source files, enforcing the execute-before-return pattern
Enable agent loops to learn from similar past tasks and share patterns across loops
Query and manage the executable feedback debug memory
Execute tests on generated code and iterate until passing