Skip to content

Implement M7.2 Explanations - queryable timelines and causal summaries #2

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
Copilot wants to merge 4 commits into from
Closed

Implement M7.2 Explanations - queryable timelines and causal summaries #2

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
d9d7151
Checkpoint from Copilot CLI for coding agent session
rgardler-msft Nov 29, 2025
78c030e
feat(explanations): implement M7.2 queryable timelines and causal sum…
Copilot Nov 29, 2025
c15e2d4
Fix magic number in timeline display, address code review feedback
Copilot Nov 29, 2025
7069ea4
Merge branch 'main' into copilot/adjacent-bison
Nov 29, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
16 changes: 16 additions & 0 deletions README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -157,6 +157,12 @@ multi-phase roadmap and `docs/gengine/how_to_play_echoes.md` for a gameplay guid
generates structured JSON reports with natural language commentary. The CLI
runner at `scripts/run_ai_observer.py` supports configurable tick budgets,
analysis intervals, and alert thresholds.
- ✅ **Phase 7 M7.2 – Explanations** shipped: queryable timelines and causal
summaries now available via `ExplanationTracker` in `gengine.echoes.sim`.
New CLI commands `timeline`, `explain`, and `why` let players query event
history, trace causal chains, and inspect agent/faction reasoning. Agent
and faction systems now capture decision context (options considered,
scores, reasoning) in their action reports for transparency.

## Repository Layout

Expand Down Expand Up @@ -404,6 +410,16 @@ seeds` block that lists which seeds attached, their target districts, and why
grab an end-of-run epilogue without exporting the entire snapshot. It pulls
from the same metadata surfaced via FastAPI `/state?detail=post-mortem` and
headless telemetry.
- `timeline [count] [scope]` – show the event timeline with optional filters.
Provide a count to limit results and optionally filter by scope (agent,
faction, environment, economy, district, system). Each event displays its
tick, message, actor, and reasoning.
- `explain <event_id>` – generate a causal explanation for a specific event.
Shows the event details, the causal chain of events that led to it, and
the actor's reasoning if available.
- `why <actor_id>` – show the reasoning history for an agent or faction.
Displays recent decisions, the options considered, and the context that
influenced each choice.
- `save <path>` – write the current snapshot as JSON.
- `load world <name>` / `load snapshot <path>` – swap to a new authored world or
on-disk snapshot (local engine mode only).
Expand Down
36 changes: 36 additions & 0 deletions docs/gengine/how_to_play_echoes.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -68,6 +68,9 @@ ASCII output as the local shell and honors `--script` for CI-friendly runs.
| `focus [district\|clear]` | Shows the current focus ring (district plus prioritized neighbors) or retargets it. The focus manager allocates more narrative budget to the selected ring; use `focus clear` to fall back to the default rotation. |
| `history [count]` | Prints the ranked narrator history (latest entries first). Each entry shows the focus center, suppressed count, and the top scored archived beats; provide an optional count to limit how many entries are shown. |
| `postmortem` | Prints a deterministic recap: environment trend deltas, the three largest faction legitimacy swings, the latest director events, and the most recent story seed states. Use it before quitting to capture an epilogue without exporting the entire snapshot. |
| `timeline [count] [scope]` | Shows the event timeline with optional filters. Provide a count to limit results and optionally filter by scope (agent, faction, environment, economy, district, system). Each event displays tick, message, actor, and reasoning. |
| `explain <event_id>` | Generates a causal explanation for a specific event. Shows event details, the causal chain of events that led to it, and actor reasoning if available. |
| `why <actor_id>` | Shows reasoning history for an agent or faction. Displays recent decisions, options considered, and the context that influenced each choice. |
| `save <path>` | Writes the current `GameState` snapshot to disk as JSON. |
| `load world <name>` | Reloads an authored world from `content/worlds/<name>/world.yml` (local engine mode only). |
| `load snapshot <path>` | Restores state from a JSON snapshot created via `save` (local engine mode only). |
Expand All @@ -76,6 +79,39 @@ ASCII output as the local shell and honors `--script` for CI-friendly runs.
Command arguments are whitespace-separated; wrap file paths containing spaces in
quotes. The shell ignores blank lines and repeats the prompt after each command.

### Explanation Commands

The `timeline`, `explain`, and `why` commands provide insight into the simulation's
decision-making process:

- **timeline**: Shows a chronological list of events with their causes and effects.
Filter by scope to focus on specific systems (e.g., `timeline 10 agent` shows
the last 10 agent actions).

- **explain**: Given an event ID from the timeline, traces back through the causal
chain to show what led to that event. Useful for understanding why a crisis
started or how faction conflicts escalated.

- **why**: Shows the reasoning behind an actor's recent decisions. For agents,
this includes district conditions and trait influences. For factions, it shows
legitimacy gaps, resource pressure, and rival analysis.

Example workflow:
```
(echoes) run 10
(echoes) timeline 5
[tick 9] agent-9-1
Aria Volt inspects Industrial Tier
why: pollution concern in Industrial Tier (pollution=0.65)
(echoes) why aria-volt
[tick 9] Aria Volt (agent)
decision: INSPECT_DISTRICT (score: 0.85)
options considered:
- INSPECT_DISTRICT: 0.85
- STABILIZE_UNREST: 0.42
- SUPPORT_SECURITY: 0.31
```

## 3. Simulation Concepts

The CLI now routes every command through the shared `SimEngine` abstraction or,
Expand Down
Loading