microsoft · Jason Robert (jrob5756) · Aug 14, 2026 · Aug 11, 2026 · Aug 12, 2026 · Aug 12, 2026
diff --git a/AGENTS.md b/AGENTS.md
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -417,6 +417,19 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
   local user can write to.
 - **Registry TOML values are escaped** — a registry whose source or type
   contained a quote or a backslash produced a file that could not be re-read.
+- **`conductor stop` against a foreground run is no longer a silent no-op.**
+  With the interactive keyboard listener active, the `SIGTERM` handler
+  delegated to the previous disposition only when it was *callable* — and in
+  an unmodified process `signal.getsignal(SIGTERM)` returns `SIG_DFL`, an
+  `IntEnum` member that is not callable, so the signal fell through and was
+  swallowed entirely: the process survived and kept running. The handler now
+  restores the default disposition and re-raises against itself. An inherited
+  `SIG_IGN` is honoured rather than converted into a termination.
+- **A `questions` node no longer leaves the run parked at an already-answered
+  gate.** A questions node reuses `gate_presented` but never emitted the
+  matching `gate_resolved`, so every consumer of the event stream — the web
+  dashboard as well as the Fleet Manager — held a gate that never closed for
+  the remainder of the run.
 - **Dashboard context-window bar no longer reports cumulative input tokens as
   a false red at >100% of the cap** (#412). The bar reused
   `AgentOutput.input_tokens` — a *billing* total summed across every API call
@@ -649,6 +662,67 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
   pane; the view is not refit in either case. The same compensation steadies
   the graph when a running workflow's topology grows — a `for_each` fanning
   out or a subworkflow's DAG arriving.
+- **`workflow_started` now records the run's resolved `inputs`.** Two runs of
+  the same workflow are otherwise indistinguishable in a listing. The values
+  are written to the run's JSONL event log, which is also read by `conductor
+  replay` and the dashboard.
+- **Fleet Manager: `conductor stop`, `conductor fleet list`, and a new
+  interactive `conductor fleet` TUI now discover every run, not just
+  `--web-bg` ones.** Previously only `--web-bg` wrote a discoverable
+  (port-keyed `.pid`) record, so a plain `conductor run` or `conductor run
+  --web` process was invisible to `conductor stop` and had to be killed by
+  hand. Every run path now writes a `run_id`-keyed JSON record to
+  `~/.conductor/runs/<run_id>.json` describing its mode (`fg`/`fg-web`/`bg`),
+  PID, workflow path, and dashboard port (when it has one); `stop`,
+  `fleet list`, and the TUI all read from this same store. The legacy
+  port-keyed `.pid` file is still read (and cleaned up) for a still-running
+  pre-upgrade process, but is no longer written by any current code path.
+  **Behavior change:** stopping a **foreground** run (`mode` `fg`/`fg-web` —
+  anything holding a terminal) now requires interactive confirmation, since a
+  plain `SIGTERM` discards in-flight progress unless periodic checkpoints are
+  enabled for that run; a background-only fleet is unaffected. Use
+  `--yes`/`-y` to skip the prompt (e.g. scripts, CI); a non-interactive
+  `stdin` without `--yes` refuses to proceed rather than silently defaulting
+  to "yes". `stop` also gained `--run-id`, the only selector that can target a
+  foreground run with no dashboard port to match on. See
+  [`docs/cli-reference.md`](docs/cli-reference.md#conductor-stop).
+
+- **`conductor fleet`** — an optional interactive Textual TUI (`pip install
+  'conductor-cli[tui]'`) for monitoring, managing, and launching Conductor
+  runs across dedicated screens: Runs (home, ~2s-polled, sorted by recency),
+  Run detail (per-agent topology and timings, not a DAG), Providers
+  (collapsed-by-default provider/model diagnostics, reusing
+  `providers/diagnostics.py`), Registries (registries → workflows → inputs),
+  New Run (form generated from a workflow's declared `input:`, launches via
+  the same `conductor run --web-bg` path the CLI uses), and History
+  (every retained run regardless of outcome, bounded by retention plus an
+  independent 200-entry display cap, delegating replay to `conductor
+  replay <log>` rather than re-implementing it). A human gate is displayed as
+  a persistent badge for every run mode; it can additionally be **resolved**
+  from the TUI (`g`) for any run with a dashboard port (`fg-web`/`bg`) via the
+  existing `conductor gate respond` HTTP path — a plain foreground run's gate
+  is display-only (its PID is shown) since its blocking prompt thread cannot
+  be reached remotely. A terminal bell / OSC 9 notification fires once per
+  transition into `at-gate` or a failure. `conductor fleet list` and
+  `conductor fleet prune` need no optional dependency; only the bare,
+  no-subcommand `conductor fleet` (which launches the TUI) requires the `tui`
+  extra. See [`docs/fleet.md`](docs/fleet.md).
+
+- **`~/.conductor/config.toml`** — a new machine-wide, read-only-in-v1
+  settings file (`src/conductor/settings.py`), read with stdlib `tomllib` and
+  honoring `$CONDUCTOR_HOME` the same way `registries.toml` does. Currently
+  controls `[fleet.retention]`: an opportunistic sweep (`enabled = true` by
+  default, `keep_last = 200`) that bounds the otherwise-unbounded
+  `$TMPDIR/conductor/` directory of event logs at the start of every
+  `conductor run`/`resume`. Never deletes the `checkpoints/` subdirectory or
+  an event log a live/resuming run still references. `conductor fleet prune`
+  is the explicit manual entry point (with `--keep-last`/`--dry-run`) and
+  always works regardless of the `enabled` setting. A missing file is normal
+  (every setting defaults cleanly); a malformed file only breaks an explicit
+  reader (`fleet prune` with no `--keep-last` override) — never `conductor
+  run`/`resume`, which swallow a settings load failure and just skip the
+  feature it configures. See
+  [`docs/configuration.md`](docs/configuration.md#machine-wide-settings-conductorconfigtoml).
 
 ## [0.1.27](https://github.com/microsoft/conductor/compare/v0.1.26...v0.1.27) - 2026-08-04
 

diff --git a/README.md b/README.md
@@ -29,6 +29,7 @@ Conductor makes multi-agent workflows — code review pipelines, research-then-s
 - **Human-in-the-loop** - Pause for human decisions with Markdown-rendered prompts and clickable file links
 - **Safety limits** - Max iterations and timeout enforcement
 - **[Web dashboard](#web-dashboard)** - Real-time workflow visualization with interactive DAG graph, breadcrumb navigation into sub-workflows, live streaming, and in-browser human gates
+- **[Fleet Manager](#fleet-manager-tui)** - An interactive TUI over every running `conductor` process (foreground, `--web`, or `--web-bg`): live status, tokens and cost, gate alerts you can answer, step-level drill-down, and launching new runs — plus non-interactive `conductor stop` / `conductor fleet list`
 - **Validation** - Catches stale template references, missing inputs, and undeclared dependencies before runtime
 
 ## Installation
@@ -211,6 +212,40 @@ conductor guide --text "Prefer Python 3.12 examples"
 conductor stop
 ```
 
+## Fleet Manager (TUI)
+
+The dashboard shows you one run in depth. The **Fleet Manager** shows you *every* run at once — and it's where you go when something needs you. Launch it with `conductor fleet`:
+
+```bash
+pip install 'conductor-cli[tui]'   # one-time: the TUI ships as an optional extra
+conductor fleet
+```
+
+![Fleet Manager](docs/img/fleet-manager.png)
+
+> **TUI = breadth. Dashboard = depth.**
+>
+> The TUI answers *"what's happening across my fleet, and what needs me?"* The dashboard answers *"what exactly is this one run doing?"* They compose — press `w` on any run to open its dashboard in a browser.
+
+**Key features:**
+
+- **Every run is discoverable** — Foreground, `--web`, and `--web-bg` runs all appear. Previously only `--web-bg` runs were visible to `conductor stop`; a plain `conductor run` had to be hunted down and killed by hand.
+- **Live fleet table** — Each run's status, current step, elapsed time, tokens, cost, and a token-burn sparkline, polled continuously and sorted by recency
+- **Gates that find you** — A run blocked on a human gate is badged in the table and fires a terminal bell, so a waiting workflow doesn't sit unnoticed. Press `g` to answer it without leaving the TUI.
+- **Drill down** — `enter` opens a run's per-agent breakdown, then `enter` again on any step shows what it actually did: its input, output, and activity stream
+- **Launch new runs** — `n` builds a form from a workflow's declared `input:` block and starts it in the background, so the TUI both watches and starts work
+- **Browse and re-run** — Providers and model diagnostics (`p`), registries and their workflows (`r`), and History (`h`) for finished runs, which hands off to `conductor replay`
+- **Kill safely** — `k` stops the selected run, `K` the whole fleet. Both confirm first, and a foreground run is named explicitly, since stopping one discards in-flight progress unless periodic checkpoints are enabled.
+
+```bash
+# Not interactive? These need no extra dependency:
+conductor fleet list           # table of every live run
+conductor stop                 # stop the only running workflow, or list them
+conductor fleet prune          # bound the event logs in $TMPDIR/conductor
+```
+
+See [docs/fleet.md](docs/fleet.md) for every screen, key binding, the status vocabulary, and retention settings.
+
 ## Providers
 
 Conductor supports multiple AI providers. Choose based on your needs:
@@ -352,6 +387,21 @@ Validate a workflow file without executing.
 conductor validate <workflow.yaml>
 ```
 
+### `conductor fleet`
+
+Discover and manage every running `conductor` process — foreground,
+`--web`, or `--web-bg`. See [Fleet Manager](#fleet-manager-tui) above for
+the interactive TUI; these need no extra dependency:
+
+```bash
+conductor stop                # stop the only running workflow, or list them
+conductor fleet list          # non-interactive table of every live run
+conductor fleet               # interactive TUI (requires the `tui` extra)
+```
+
+See [docs/fleet.md](docs/fleet.md) for the TUI's screens, key bindings, and
+status vocabulary.
+
 **Full CLI documentation:** [docs/cli-reference.md](docs/cli-reference.md)
 
 ## Workflow Registries
@@ -401,6 +451,7 @@ See the [`examples/`](./examples/) directory for complete workflows:
 |----------|-------------|
 | [Workflow Syntax](./docs/workflow-syntax.md) | Complete YAML schema reference |
 | [CLI Reference](./docs/cli-reference.md) | Full command-line documentation |
+| [Fleet Manager](./docs/fleet.md) | `conductor fleet` TUI: screens, key bindings, gate resolvability, retention |
 | [Parallel Execution](./docs/parallel-execution.md) | Static parallel groups |
 | [Dynamic Parallel](./docs/dynamic-parallel.md) | For-each groups and array processing |
 | [Claude Provider](./docs/providers/claude.md) | Claude setup and configuration |