agent-bios 0.9.1 → 0.9.3

package/claude/guides/claude-prompting.md

@@ -17,7 +17,7 @@ core_rules:
   - name the boundary explicitly — what to do without asking, and what to stop and ask about
 targets:
   - claude-fable-5
-  - claude-opus-4-8
+  - claude-opus-5
   - claude-sonnet-5
   - claude-haiku-4-5
 verification_focus:
@@ -55,15 +55,21 @@ pick the path.
 ## When to add blocks
 
 - Long-horizon or autonomous work: give the full spec up front in one
-  well-specified turn and run at a high effort. Add an explicit self-check
-  cadence; separate fresh-context verifier subagents beat self-critique.
+  well-specified turn and run at a high effort. Do **not** add a self-check
+  cadence or a dedicated verifier subagent: the helm binding verifies its own
+  work unasked, so an instruction to verify buys over-verification instead.
+  Deleting inherited verification scaffolding costs no capability — this inverts
+  the usual self-check advice, so carve this tier out of a prompt library that
+  applies that advice uniformly.
 - Review: state the evidence bar and the verdict shape. This tier follows
   severity filters literally, so "only report high-severity" depresses measured
   recall even as bug-finding improves — ask for every finding with confidence
   and severity attached, and filter downstream.
-- Delegation: say when to delegate. Left unprompted this tier under-reaches for
-  subagents, file-based memory, and custom tools — it will not spend an
-  expensive capability unless it is reasonably sure the capability is needed.
+- Delegation: say when *not* to delegate, and cap the spawn count. The helm
+  binding reaches for subagents readily — the reverse of the binding it replaced
+  — and every spawn rebuilds context, reports back, and is then re-read, so
+  unbounded delegation multiplies cost and latency. File-based memory and custom
+  tools are the opposite case: they still need an explicit when-to-use trigger.
 - Autonomous runs with no human watching: say so. Otherwise it asks permission
   it does not need and blocks. Grant autonomy on minor choices (naming,
   defaults, equivalent approaches) while keeping the ask for scope changes and
@@ -80,13 +86,17 @@ pick the path.
   per subtask, and the orchestrator is not pinned to the slowest one.
 - Effort ladder — `low`, `medium`, `high`, `xhigh`, `max` on the frontier, helm,
   and workhorse bindings; `xhigh` is the best setting for most coding and
-  agentic work, `high` a sound default, `low` for cheap subagent scans. Sweep the
-  levels on a real eval set rather than reaching for the top: higher effort up
-  front often *reduces* turn count and total cost on agentic work, while some
-  tasks land equally well a level down.
+  agentic work, `high` a sound default, `low` for cheap subagent scans. Start
+  there and sweep *down* on a real eval set rather than reaching for the top: the
+  lower rungs are unusually strong on the helm binding, higher effort up front
+  often *reduces* turn count and total cost on agentic work, and some tasks land
+  equally well a level down. Effort does not control response length — see
+  Working rules.
 - Per-model constraints differ across the `targets` bindings — thinking
   configuration, sampling parameters, and effort support are not uniform, and
-  the sweep binding is the most restricted. Confirm the constraint against the
+  the sweep binding is the most restricted. On the helm binding, for instance,
+  thinking is on by default and turning it off is accepted only at `high` effort
+  or below. Confirm the constraint against the
   live surface before relying on it in a dispatch; do not assume the frontier
   binding's rules apply to the sweep one.
 
@@ -98,6 +108,13 @@ pick the path.
 - Do not add "summarize every N tool calls" scaffolding — this tier narrates on
   its own. If it narrates too much for a coding agent, set a silence default
   instead: text only on a finding, a direction change, or a blocker.
+- Length is a prompting lever, not an effort lever. This tier writes longer
+  answers and longer files than its predecessors, and lowering `effort` does not
+  reliably shorten visible output — only an explicit instruction does. Calibrate
+  the deliverable's length separately from the conversation's.
+- Scope self-correction. Left alone this tier narrates its own earlier mistakes
+  at length, which reads as thrash. Ask it to correct only what would change the
+  reader's decisions, say it plainly, and carry on.
 - Give it somewhere to write learnings, tell it to consult that place later, and
   give the file a format. It performs notably better with a memory surface.
 - Keep the deliverable readable: the final message is the reader's first look at
@@ -111,7 +128,8 @@ pick the path.
 2. Name the boundaries — what to do freely, what to stop and ask about.
 3. Give each tool a when-to-call description.
 4. Say how progress claims must be grounded, and how the deliverable should read.
-5. Delete the step-by-step scaffolding inherited from older models, then A/B the
+5. Delete the step-by-step scaffolding inherited from older models — including
+   any verify-your-work step and any encouragement to delegate — then A/B the
    prompt with it removed before keeping either version.
 
 ## Sources

package/claude/guides/cli-multi-model-workflow.md

@@ -156,12 +156,12 @@ Write for the next agent and re-verification, not narrative. Required content:
 
 This is the human-readable projection of concrete models/tools; `config/agent-launch.toml` is the machine launch authority and parity checks keep them aligned. Re-probe when the binding is older than ~8 weeks or a newer observable model/tool changes the surface. `agent-bios install` overwrites deployed bindings, so edit the repo copy.
 
-Binding (2026-07-13):
+Binding (2026-07-25):
 
 | Slot | Binding | Notes |
 |---|---|---|
 | FRONTIER | Claude Fable 5 · GPT-5.6 Sol (read-only; max default, task-fit effort including Ultra) | bounded hardest decisions and verdicts |
-| HELM | Claude Opus 4.8 (xhigh) · GPT-5.6 Sol (xhigh main; main Ultra requires explicit selection; bounded FRONTIER Ultra allowed) | standing main; Codex defaults bypass, explicit sandbox narrows |
+| HELM | Claude Opus 5 (xhigh) · GPT-5.6 Sol (xhigh main; main Ultra requires explicit selection; bounded FRONTIER Ultra allowed) | standing main; Codex defaults bypass, explicit sandbox narrows |
 | WORKHORSE | Claude Sonnet 5 · GPT-5.6 Terra (high) | implementation and per-item judgment |
 | SWEEP | Claude Haiku 4.5 · GPT-5.6 Luna (low) | clear repeatable scans and mechanical work |
 | VERIFIER-A | onto-mcp review | logic/authority kind; tier units per funnel |
@@ -186,7 +186,7 @@ Dispatch packets:
 | GPT-5.6 Sol FRONTIER | outcome, evidence, decision boundary, stop/verification; task-fit effort; read-only |
 | GPT-5.6 Terra WORKHORSE | outcome, frozen scope/inputs, authority, done-when, evidence/report, escalation; high |
 | GPT-5.6 Luna SWEEP | exact search space, rules, ambiguity behavior, stop, output; low; no architecture/debugging |
-| Claude Opus 4.8 HELM | xhigh for agentic work; high minimum for sensitive judgment; lower only when bounded/cost-led |
+| Claude Opus 5 HELM | xhigh for agentic work; high minimum for sensitive judgment; lower only when bounded/cost-led |
 | Claude Sonnet 5 WORKHORSE | exact scope, apply-to-all rules, tools, verification, report; high default |
 | Claude Haiku 4.5 SWEEP | closed, high-volume, latency-sensitive work with exact schema/examples |
 

package/codex/guides/claude-prompting.md

@@ -17,7 +17,7 @@ core_rules:
   - name the boundary explicitly — what to do without asking, and what to stop and ask about
 targets:
   - claude-fable-5
-  - claude-opus-4-8
+  - claude-opus-5
   - claude-sonnet-5
   - claude-haiku-4-5
 verification_focus:
@@ -55,15 +55,21 @@ pick the path.
 ## When to add blocks
 
 - Long-horizon or autonomous work: give the full spec up front in one
-  well-specified turn and run at a high effort. Add an explicit self-check
-  cadence; separate fresh-context verifier subagents beat self-critique.
+  well-specified turn and run at a high effort. Do **not** add a self-check
+  cadence or a dedicated verifier subagent: the helm binding verifies its own
+  work unasked, so an instruction to verify buys over-verification instead.
+  Deleting inherited verification scaffolding costs no capability — this inverts
+  the usual self-check advice, so carve this tier out of a prompt library that
+  applies that advice uniformly.
 - Review: state the evidence bar and the verdict shape. This tier follows
   severity filters literally, so "only report high-severity" depresses measured
   recall even as bug-finding improves — ask for every finding with confidence
   and severity attached, and filter downstream.
-- Delegation: say when to delegate. Left unprompted this tier under-reaches for
-  subagents, file-based memory, and custom tools — it will not spend an
-  expensive capability unless it is reasonably sure the capability is needed.
+- Delegation: say when *not* to delegate, and cap the spawn count. The helm
+  binding reaches for subagents readily — the reverse of the binding it replaced
+  — and every spawn rebuilds context, reports back, and is then re-read, so
+  unbounded delegation multiplies cost and latency. File-based memory and custom
+  tools are the opposite case: they still need an explicit when-to-use trigger.
 - Autonomous runs with no human watching: say so. Otherwise it asks permission
   it does not need and blocks. Grant autonomy on minor choices (naming,
   defaults, equivalent approaches) while keeping the ask for scope changes and
@@ -80,13 +86,17 @@ pick the path.
   per subtask, and the orchestrator is not pinned to the slowest one.
 - Effort ladder — `low`, `medium`, `high`, `xhigh`, `max` on the frontier, helm,
   and workhorse bindings; `xhigh` is the best setting for most coding and
-  agentic work, `high` a sound default, `low` for cheap subagent scans. Sweep the
-  levels on a real eval set rather than reaching for the top: higher effort up
-  front often *reduces* turn count and total cost on agentic work, while some
-  tasks land equally well a level down.
+  agentic work, `high` a sound default, `low` for cheap subagent scans. Start
+  there and sweep *down* on a real eval set rather than reaching for the top: the
+  lower rungs are unusually strong on the helm binding, higher effort up front
+  often *reduces* turn count and total cost on agentic work, and some tasks land
+  equally well a level down. Effort does not control response length — see
+  Working rules.
 - Per-model constraints differ across the `targets` bindings — thinking
   configuration, sampling parameters, and effort support are not uniform, and
-  the sweep binding is the most restricted. Confirm the constraint against the
+  the sweep binding is the most restricted. On the helm binding, for instance,
+  thinking is on by default and turning it off is accepted only at `high` effort
+  or below. Confirm the constraint against the
   live surface before relying on it in a dispatch; do not assume the frontier
   binding's rules apply to the sweep one.
 
@@ -98,6 +108,13 @@ pick the path.
 - Do not add "summarize every N tool calls" scaffolding — this tier narrates on
   its own. If it narrates too much for a coding agent, set a silence default
   instead: text only on a finding, a direction change, or a blocker.
+- Length is a prompting lever, not an effort lever. This tier writes longer
+  answers and longer files than its predecessors, and lowering `effort` does not
+  reliably shorten visible output — only an explicit instruction does. Calibrate
+  the deliverable's length separately from the conversation's.
+- Scope self-correction. Left alone this tier narrates its own earlier mistakes
+  at length, which reads as thrash. Ask it to correct only what would change the
+  reader's decisions, say it plainly, and carry on.
 - Give it somewhere to write learnings, tell it to consult that place later, and
   give the file a format. It performs notably better with a memory surface.
 - Keep the deliverable readable: the final message is the reader's first look at
@@ -111,7 +128,8 @@ pick the path.
 2. Name the boundaries — what to do freely, what to stop and ask about.
 3. Give each tool a when-to-call description.
 4. Say how progress claims must be grounded, and how the deliverable should read.
-5. Delete the step-by-step scaffolding inherited from older models, then A/B the
+5. Delete the step-by-step scaffolding inherited from older models — including
+   any verify-your-work step and any encouragement to delegate — then A/B the
    prompt with it removed before keeping either version.
 
 ## Sources

package/codex/guides/cli-multi-model-workflow.md

@@ -156,12 +156,12 @@ Write for the next agent and re-verification, not narrative. Required content:
 
 This is the human-readable projection of concrete models/tools; `config/agent-launch.toml` is the machine launch authority and parity checks keep them aligned. Re-probe when the binding is older than ~8 weeks or a newer observable model/tool changes the surface. `agent-bios install` overwrites deployed bindings, so edit the repo copy.
 
-Binding (2026-07-13):
+Binding (2026-07-25):
 
 | Slot | Binding | Notes |
 |---|---|---|
 | FRONTIER | Claude Fable 5 · GPT-5.6 Sol (read-only; max default, task-fit effort including Ultra) | bounded hardest decisions and verdicts |
-| HELM | Claude Opus 4.8 (xhigh) · GPT-5.6 Sol (xhigh main; main Ultra requires explicit selection; bounded FRONTIER Ultra allowed) | standing main; Codex defaults bypass, explicit sandbox narrows |
+| HELM | Claude Opus 5 (xhigh) · GPT-5.6 Sol (xhigh main; main Ultra requires explicit selection; bounded FRONTIER Ultra allowed) | standing main; Codex defaults bypass, explicit sandbox narrows |
 | WORKHORSE | Claude Sonnet 5 · GPT-5.6 Terra (high) | implementation and per-item judgment |
 | SWEEP | Claude Haiku 4.5 · GPT-5.6 Luna (low) | clear repeatable scans and mechanical work |
 | VERIFIER-A | onto-mcp review | logic/authority kind; tier units per funnel |
@@ -186,7 +186,7 @@ Dispatch packets:
 | GPT-5.6 Sol FRONTIER | outcome, evidence, decision boundary, stop/verification; task-fit effort; read-only |
 | GPT-5.6 Terra WORKHORSE | outcome, frozen scope/inputs, authority, done-when, evidence/report, escalation; high |
 | GPT-5.6 Luna SWEEP | exact search space, rules, ambiguity behavior, stop, output; low; no architecture/debugging |
-| Claude Opus 4.8 HELM | xhigh for agentic work; high minimum for sensitive judgment; lower only when bounded/cost-led |
+| Claude Opus 5 HELM | xhigh for agentic work; high minimum for sensitive judgment; lower only when bounded/cost-led |
 | Claude Sonnet 5 WORKHORSE | exact scope, apply-to-all rules, tools, verification, report; high default |
 | Claude Haiku 4.5 SWEEP | closed, high-volume, latency-sensitive work with exact schema/examples |
 

package/config/agent-launch.toml

@@ -48,7 +48,7 @@ workhorse = "${CODEX_HOME}/agents/workhorse.toml"
 sweep = "${CODEX_HOME}/agents/sweep.toml"
 
 [hosts.claude]
-models = ["claude-fable-5", "claude-opus-4-8", "claude-sonnet-5", "claude-haiku-4-5"]
+models = ["claude-fable-5", "claude-opus-5", "claude-sonnet-5", "claude-haiku-4-5"]
 # onto review seat used when a CROSS-family main (Codex) routes onto review to
 # this (anthropic/claude) family; must be an onto review-role registered (provider, model).
 onto_review = { provider = "anthropic", model = "claude-fable-5" }
@@ -58,7 +58,7 @@ model = "claude-fable-5"
 effort = "max"
 
 [hosts.claude.tiers.helm]
-model = "claude-opus-4-8"
+model = "claude-opus-5"
 effort = "xhigh"
 
 [hosts.claude.tiers.workhorse]

package/package.json

@@ -1,6 +1,7 @@
 {
   "name": "agent-bios",
-  "version": "0.9.1",
+  "version": "0.9.3",
+  "releaseDate": "2026-07-25",
   "description": "A thin, low-level instruction layer for LLM CLI agents: one set of principles and behavior whichever model you run. Deploys into $HOME by copy via an explicit `agent-bios install`.",
   "bin": {
     "agent-bios": "scripts/install.sh"

package/scripts/agent-launch.py

@@ -609,6 +609,9 @@ def _build_app_class():
     def setup_panel(plan):
         panel = Static("\n".join(setup_summary_lines(plan)), id="al-setup")
         panel.border_title = "Current setup"
+        release = version_label()
+        if release:
+            panel.border_subtitle = release
         return panel
 
     class MenuScreen(ModalScreen):
@@ -1460,6 +1463,13 @@ CORPUS_STATUS_PATH = pathlib.Path(
     )
 )
 
+VERSION_INFO_PATH = pathlib.Path(
+    os.environ.get(
+        "AGENT_LAUNCH_VERSION_FILE",
+        str(pathlib.Path.home() / ".local/share/agent-bios/version.json"),
+    )
+)
+
 
 def load_corpus_status() -> dict[str, Any] | None:
     try:
@@ -1468,6 +1478,22 @@ def load_corpus_status() -> dict[str, Any] | None:
         return None
 
 
+def version_label() -> str | None:
+    """The deployed agent-bios version + release date for the TUI, or None when
+    the marker is absent (uninstalled / dev checkout). `agent-bios install`
+    writes it from package.json (version + releaseDate). This is the deploy /
+    system version — distinct from the corpus content version in the distill hub."""
+    try:
+        info = json.loads(VERSION_INFO_PATH.read_text())
+    except (OSError, ValueError):
+        return None
+    version = info.get("version") if isinstance(info, dict) else None
+    if not version:
+        return None
+    released = info.get("releaseDate")
+    return f"agent-bios v{version} · {released}" if released else f"agent-bios v{version}"
+
+
 def corpus_summary_lines(status: dict[str, Any] | None) -> list[str]:
     """Panel body for the Session Distill area: which corpus content is
     live, through which mechanisms, and whether the corpus is rolled back."""

package/scripts/check-parity.sh

@@ -273,7 +273,7 @@ for tier, (model, effort) in expected_launch_tiers.items():
 
 expected_claude_tiers = {
     "frontier": ("claude-fable-5", "max"),
-    "helm": ("claude-opus-4-8", "xhigh"),
+    "helm": ("claude-opus-5", "xhigh"),
     "workhorse": ("claude-sonnet-5", "high"),
     "sweep": ("claude-haiku-4-5", "low"),
 }
@@ -409,7 +409,7 @@ for guide_name in [
         mark_fail(f"{guide} SWEEP must bind exact GPT-5.6 Luna")
     for slot, claude_model in (
         ("FRONTIER", "Claude Fable 5"),
-        ("HELM", "Claude Opus 4.8"),
+        ("HELM", "Claude Opus 5"),
         ("WORKHORSE", "Claude Sonnet 5"),
         ("SWEEP", "Claude Haiku 4.5"),
     ):
@@ -764,6 +764,26 @@ if launcher.is_file() and launch_profile:
                 "project an applied setup (customization must not be silently discarded)"
             )
 
+        # Version line: version_label reads the deployed version marker
+        # ($STATE_DIR/version.json) that install writes from package.json. A
+        # present marker yields "agent-bios v<version> · <releaseDate>"; an absent
+        # one yields None (uninstalled / dev checkout, so the panel shows no line).
+        saved_version_path = launcher_module.VERSION_INFO_PATH
+        try:
+            version_marker = pathlib.Path(tempfile.mkdtemp()) / "version.json"
+            version_marker.write_text('{"version": "9.9.9", "releaseDate": "2026-01-02"}')
+            launcher_module.VERSION_INFO_PATH = version_marker
+            if launcher_module.version_label() != "agent-bios v9.9.9 · 2026-01-02":
+                mark_fail(
+                    "agent-launch version_label wrong with a marker present: "
+                    f"{launcher_module.version_label()!r}"
+                )
+            launcher_module.VERSION_INFO_PATH = version_marker.parent / "absent.json"
+            if launcher_module.version_label() is not None:
+                mark_fail("agent-launch version_label must be None when the marker is absent")
+        finally:
+            launcher_module.VERSION_INFO_PATH = saved_version_path
+
     with tempfile.TemporaryDirectory() as raw_tmp:
         tmp = pathlib.Path(raw_tmp)
         backend = tmp / "backend"
@@ -806,10 +826,13 @@ if launcher.is_file() and launch_profile:
             mark_fail(f"fake launch profile did not replace every backend: {fake_commands!r}")
             fake_profile = pathlib.Path("/nonexistent/unsafe-fixture")
         env = os.environ.copy()
+        version_fixture = tmp / "version.json"
+        version_fixture.write_text('{"version": "9.9.9", "releaseDate": "2026-01-02"}')
         env.update(
             FAKE_LAUNCH_ARGV=str(argv_log),
             AGENT_LAUNCH_CONFIG=str(fake_profile),
             XDG_CACHE_HOME=str(tmp / "cache"),
+            AGENT_LAUNCH_VERSION_FILE=str(version_fixture),
         )
 
         pty_env = env.copy()
@@ -974,6 +997,8 @@ if launcher.is_file() and launch_profile:
             < positions[b"Options ("]
         ):
             mark_fail("agent-launch layout is not setup then description then options")
+        if b"agent-bios v9.9.9" not in transcript:
+            mark_fail("agent-launch does not show the deployed version line in the setup panel")
 
         _, picker_status = run_picker_scenario(
             "root escape cancellation",

package/scripts/install.sh

@@ -476,6 +476,24 @@ cmd_install() {
   else
     log "note: corpus-status projection unavailable (versions.json/ledger missing?)"
   fi
+  # Deploy/system version marker for the launcher's TUI version line, read from
+  # package.json (version + releaseDate) — distinct from the corpus content
+  # version. Best-effort: a failure here never fails the install.
+  if [ "$DRY_RUN" != 1 ]; then
+    if python3 - "$REPO/package.json" "$STATE_DIR/version.json" <<'PY' 2>/dev/null
+import json, sys
+pkg = json.load(open(sys.argv[1]))
+with open(sys.argv[2], "w") as f:
+    json.dump({"version": pkg.get("version"), "releaseDate": pkg.get("releaseDate")}, f)
+    f.write("\n")
+PY
+    then
+      printf '%s\n' "$STATE_DIR/version.json" >> "$MANIFEST"
+      info "version marker written ($STATE_DIR/version.json)"
+    else
+      log "note: version marker not written (package.json unreadable)"
+    fi
+  fi
   log ""
   log "Verifying deployment..."
   if cmd_verify; then

package/scripts/session-cost.py

@@ -12,6 +12,7 @@ from datetime import datetime
 PRICES = {
     "claude-fable-5":  (10.0, 50.0, 1.00, 12.50, 20.0),
     "claude-mythos-5": (10.0, 50.0, 1.00, 12.50, 20.0),
+    "claude-opus-5":   (5.0, 25.0, 0.50, 6.25, 10.0),
     "claude-opus-4-8": (5.0, 25.0, 0.50, 6.25, 10.0),
     "claude-opus-4-7": (5.0, 25.0, 0.50, 6.25, 10.0),
     "claude-opus-4-6": (5.0, 25.0, 0.50, 6.25, 10.0),