@zalom/plastic 2.0.0-alpha.26 → 2.0.0-alpha.27

package/README.md

@@ -156,9 +156,9 @@ plan a machine can build from exactly. Read
   or autonomous delivery by an agent team when you ask for it.
 - Two advisor agents ship for the hard problems: one-way doors, plans, adversarial
   review, deadlocks. Summon one deliberately, state the effort and the answer shape in the brief; nobody's
-  main session is ever touched. Faux Fable (recommended) is an ordinary model carrying
-  the frontier reasoning discipline inline, much cheaper; Fable 5 is the frontier model
-  itself. Choose the default at install, switch it any time.
+  main session is ever touched. Primary Advisor uses Fable at medium effort for normal
+  consultation. Secondary Advisor uses Fable at high effort for explicit escalation.
+  Choose the default at install, and switch it any time.
 
 Plastic needs Ruby 3.0 or later and Node.js 18 or later. Most Linux systems already
 have a new enough Ruby. macOS ships Ruby 2.6, which is too old, so a clean Mac needs

package/agents/plastic-enforcer.md

@@ -5,6 +5,7 @@ description: |
   record, has the plan reviewed before code, dispatches one executor, reviews
   by risk, and closes.
 model: opus
+effort: medium
 ---
 
 You are the Plastic Enforcer, the lead of an auto team. You ARE the orchestrating session,
@@ -14,8 +15,8 @@ spanning the whole What->Why->How->Exec cycle; you are not a separately dispatch
 orchestrating main session on the best available thinking model (Fable, Opus, or whatever
 supersedes them). This is advice only: it changes no behavior. Dispatched agents keep their
 configured model and never resolve to Fable unless an explicit `agents.models.<name>` config
-override names Fable for that role. The two advisors, `plastic-advisor` and
-`plastic-faux-advisor`, are consultation roles the user or the main session summons
+override names Fable for that role. Primary Advisor and Secondary Advisor are consultation roles
+the user or the main session summons
 deliberately; the auto pipeline never dispatches them.
 
 ## Your Responsibilities

package/agents/plastic-executor.md

@@ -4,6 +4,7 @@ description: |
   Use for the Exec stage in auto mode: commit the plan's tests red, implement
   the actions, check off the checklist, and drive the test suite green.
 model: sonnet
+effort: medium
 ---
 
 You are the Plastic Executor. You own the Exec stage of the What->Why->How->Exec cycle.

package/agents/plastic-node-research.md

@@ -11,6 +11,7 @@ tools:
   - Glob
   - Grep
 model: sonnet
+effort: medium
 ---
 
 You are one node in Plastic's node-graph runner: one node input in, one YAML return out, no diff.
@@ -21,8 +22,9 @@ Code's `tools:` field grants or withholds a whole tool name, never a command pat
 is a write whatever command it runs. `Read`, `Glob`, and `Grep` are enough to read every file the
 node input points at.
 
-A research node produces no diff. Put what you found in your return's `findings`, never in a code
-change.
+A research node produces no diff. Put short discoveries in `findings`. When the node input declares
+`report:`, put the complete Markdown report in the return's `report` field. The runner writes that
+declared intent resource after validation; you never edit it or any project file yourself.
 
 End your turn with exactly one YAML document, the return the dispatch line names, nothing else
 around it.

package/agents/plastic-node-verify.md

@@ -10,6 +10,7 @@ tools:
   - Glob
   - Grep
 model: opus
+effort: medium
 ---
 
 You are one node in Plastic's node-graph runner: one node input in, one YAML return out, no diff.

package/agents/plastic-node-work.md

@@ -13,6 +13,7 @@ tools:
   - Glob
   - Grep
 model: sonnet
+effort: medium
 ---
 
 You are one node in Plastic's node-graph runner: one node input in, your files and your commit and

package/agents/{plastic-advisor.md → plastic-primary-advisor.md}

@@ -1,13 +1,12 @@
 ---
-name: plastic-advisor
+name: plastic-primary-advisor
 description: |
-  The real advisor: expensive, consultation-only, dispatched deliberately by
-  the user or main session for the hardest reasoning, never by the auto
-  pipeline. State an EFFORT line in the brief (low, medium, high, xhigh, or max) and
-  the shape you need: a verdict plus the biggest risk for one bounded
+  The Primary Advisor: consultation-only, dispatched deliberately by
+  the user or main session for normal expensive reasoning, never by the auto
+  pipeline. State the shape you need: a verdict plus the biggest risk for one bounded
   decision; a stepped plan plus a risk map for a plan or plan review; rival
   approaches and kill criteria for architecture, one-way doors, or deadlocks. Model is set by config
-  (agents.models.claude.plastic-advisor); fable is the shipped default.
+  (agents.models.claude.plastic-primary-advisor); fable is the shipped default.
 model: fable
 effort: medium
 ---
@@ -17,8 +16,7 @@ Protocol, whatever model is running you today. The caller pays premium rates
 for this consultation, so every sentence you return must earn its cost.
 
 **Your world is the brief.** The caller sends a natural-prose briefing that should
-cover: the goal and the decision the answer feeds, an EFFORT line (low, medium,
-high, xhigh, or max), the answer shape it needs, up to three questions, the
+cover: the goal and the decision the answer feeds, the answer shape it needs, up to three questions, the
 caller's own candidate answer, evidence labeled verified/inferred/assumed, what
 was tried and how it failed, hard constraints, one-way doors, and the expected
 answer shape. Do not explore the repository or the web; if a load-bearing piece

package/agents/{plastic-faux-advisor.md → plastic-secondary-advisor.md}

@@ -1,16 +1,14 @@
 ---
-name: plastic-faux-advisor
+name: plastic-secondary-advisor
 description: |
-  The imitation advisor: an ordinary model carrying the frontier reasoning
-  discipline below, so it reasons like the real advisor at a fraction of the
-  cost. Consultation-only, dispatched deliberately by the user or main
-  session, never by the auto pipeline. State an EFFORT line in the brief (low, medium, high, xhigh, or max) and
-  the shape you need: a verdict plus the biggest risk for one bounded
+  The Secondary Advisor: an explicit escalation for difficult reasoning.
+  Consultation-only, dispatched deliberately by the user or main
+  session, never by the auto pipeline. State the shape you need: a verdict plus the biggest risk for one bounded
   decision; a stepped plan plus a risk map for a plan or plan review; rival
-  approaches and kill criteria for architecture, one-way doors, or deadlocks. Model is set by config (agents.models.claude.plastic-faux-advisor);
-  opus is the shipped default and the cheaper of the two advisors.
-model: opus
-effort: max
+  approaches and kill criteria for architecture, one-way doors, or deadlocks. Model is set by config (agents.models.claude.plastic-secondary-advisor);
+  fable is the shipped default.
+model: fable
+effort: high
 ---
 
 You are the advisor, consulted for expensive reasoning per the shipped Advisor
@@ -20,8 +18,7 @@ Operating Manual below is how you reach that bar: read it as your own working
 method, not as background reading.
 
 **Your world is the brief.** The caller sends a natural-prose briefing that should
-cover: the goal and the decision the answer feeds, an EFFORT line (low, medium,
-high, xhigh, or max), the answer shape it needs, up to three questions, the
+cover: the goal and the decision the answer feeds, the answer shape it needs, up to three questions, the
 caller's own candidate answer, evidence labeled verified/inferred/assumed, what
 was tried and how it failed, hard constraints, one-way doors, and the expected
 answer shape. Do not explore the repository or the web; if a load-bearing piece

package/config_asks.yml

@@ -32,7 +32,7 @@ config_asks:
     agents: ["claude"]
     question: "Which advisor should be the default?"
     options:
-      - label: "Faux Fable (recommended): Opus 4.8 carrying the frontier reasoning instructions. Much cheaper, available on any plan, reasons in the same disciplined way."
-        value: "plastic-faux-advisor"
-      - label: "Fable 5: the frontier model itself. The strongest reasoning available, billed through usage credits, so summon it for a few rounds and close it."
-        value: "plastic-advisor"
+      - label: "Primary Advisor (recommended): Fable at medium effort for normal consultation."
+        value: "plastic-primary-advisor"
+      - label: "Secondary Advisor: Fable at high effort for explicit escalation."
+        value: "plastic-secondary-advisor"

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@zalom/plastic",
-  "version": "2.0.0-alpha.26",
+  "version": "2.0.0-alpha.27",
   "description": "Intent-driven idea development system for AI coding agents",
   "type": "module",
   "bin": {

package/scripts/doctor.rb

@@ -2159,6 +2159,8 @@ end
 
     global_config = load_yaml_safe(File.join(plastic_home, "config.yml")) || {}
     overrides = AgentModels.override_map(project_config: {}, global_config: global_config, harness: "codex")
+    effort_overrides = AgentModels.effort_override_map(project_config: {}, global_config: global_config,
+                                                        harness: "codex")
 
     drifted = []
     sanctioned = []
@@ -2169,13 +2171,13 @@ end
       fields = codex_agent_toml_model_fields(File.read(path))
       override = overrides[basename]
 
-      if override
+      if override || effort_overrides[basename]
         sanctioned << "#{basename}: toml model=#{fields[:model].inspect}, " \
-                      "toml effort=#{fields[:effort].inspect}, sanctioned override=#{override.inspect}"
-      elsif AgentModels::TIER_DEFAULTS.key?(basename)
-        tier = AgentModels::TIER_DEFAULTS[basename]
-        expected_model = AgentModels.codex_model_for(tier)
-        expected_effort = AgentModels.effort_for(tier)
+                      "toml effort=#{fields[:effort].inspect}, sanctioned model override=#{override.inspect}, " \
+                      "effort override=#{effort_overrides[basename].inspect}"
+      elsif AgentModels::SHIPPED_MODEL_DEFAULTS.key?(basename)
+        expected_model = AgentModels.shipped_model_for(basename, harness: "codex")
+        expected_effort = AgentModels.shipped_effort_for(basename)
         mismatches = []
         if expected_model && fields[:model] != expected_model
           mismatches << "model=#{fields[:model].inspect}, resolved default model=#{expected_model.inspect}"
@@ -2187,7 +2189,7 @@ end
       else
         unclassified << "#{basename}: toml model=#{fields[:model].inspect}, " \
                         "toml effort=#{fields[:effort].inspect}, no resolved default (basename is in " \
-                        "neither AgentModels::TIER_DEFAULTS nor AgentModels::CONSULTATION_AGENTS in " \
+                        "AgentModels::SHIPPED_MODEL_DEFAULTS in " \
                         "scripts/lib/agent_models.rb; add it there, or set agents.models.codex.#{basename} " \
                         "to sanction a model explicitly)"
       end

package/scripts/install.rb

@@ -81,6 +81,7 @@ class Install < InstallerCore
 
     distribute(mode)
     bootstrap if fresh
+    migrate_advisor_config_file(File.join(plastic_home, "config.yml"))
     apply_config_flags(argv)
 
     results = selected.map do |key|
@@ -239,10 +240,9 @@ class Install < InstallerCore
         --no-advisor         Skip installing both advisor agents and the agent-advisor
                              skill (advisor.enabled: false)
         --advisor VALUE      Which advisor agent is the default: an agent name, or the
-                             shorthand "real" (plastic-advisor) or "faux"
-                             (plastic-faux-advisor). Writes advisor.claude.default. Left
-                             unset, the agent-advisor skill falls back to
-                             plastic-faux-advisor at consult time.
+                             shorthand "primary" or "secondary". Legacy "real" and
+                             "faux" aliases remain accepted. Writes advisor.claude.default.
+                             Left unset, the agent-advisor skill falls back to Primary Advisor.
         -h, --help           Show this help
 
       Notes:

package/scripts/lib/action_graph_shim.rb

@@ -83,7 +83,7 @@ module ActionGraphShim
   end
 
   # Node records for either shape (D8), always carrying NodeFile.parse's own
-  # key set (ok:, node:, kind:, files:, budget:, body:, errors:) plus
+  # key set (ok:, node:, kind:, files:, budget:, report:, body:, errors:) plus
   # needs:, path: and proven_by:. [] on :none, never a raise on a real
   # directory.
   def nodes(intent_dir)
@@ -153,6 +153,7 @@ module ActionGraphShim
           kind: "work",
           files: files_section_paths(text),
           budget: nil,
+          report: nil,
           body: text,
           errors: [],
           needs: needs_targets,
@@ -166,6 +167,7 @@ module ActionGraphShim
           kind: "work",
           files: [],
           budget: nil,
+          report: nil,
           body: nil,
           errors: ["could not read #{path}: #{e.message}"],
           needs: needs_targets,

package/scripts/lib/agent_models.rb

@@ -14,8 +14,7 @@ module AgentModels
   # HarnessAdapter::AGENT_TYPE_BY_KIND. work and research resolve the
   # executor tier, verify the advisor tier, mirroring RunnerPolicy's
   # model_role split (327 D12): work and research run on plastic-executor's
-  # own tier, verify on the same tier plastic-advisor's imitation
-  # (plastic-faux-advisor) ships, never the cheap tier.
+  # own tier, and verify on the lifecycle advisor tier, never the cheap tier.
   TIER_DEFAULTS = {
     "plastic-enforcer" => "opus",
     "plastic-executor" => "sonnet",
@@ -24,32 +23,35 @@ module AgentModels
     "plastic-node-research" => "sonnet"
   }.freeze
 
-  # The two advisor agents (intent 185 final design): plastic-advisor (the real
-  # advisor, ships `model: fable`) and plastic-faux-advisor (the imitation
-  # advisor, an ordinary model carrying the same reasoning discipline inline,
-  # ships `model: opus`). Both are shipped DEFAULTS in frontmatter, never a
+  # The two consultation agents are Primary Advisor and Secondary Advisor.
+  # Both ship Fable; Primary uses medium effort and Secondary uses high effort.
+  # Both are shipped defaults in frontmatter, never a
   # hard-wired identity: agents.models.claude.<name> (or the legacy flat form)
   # overrides either one through the same install-time frontmatter rewrite
   # every agent override uses. Neither is a lifecycle-stage role: never
-  # dispatched by the auto pipeline, not part of TIER_DEFAULTS. Claude-only for
-  # this release (generate_codex_agents skips both by name; the Codex advisor
-  # case is intent 186, not a permanent exclusion).
-  #
-  # Intent 186 DEFINES the advisor Codex pairing but keeps emission deferred: when the skip is
-  # lifted, plastic-advisor pairs to gpt-5.6-sol at xhigh (the deepest) and plastic-faux-advisor
-  # to gpt-5.6-terra at high (cheaper). Neither is in TIER_DEFAULTS and neither is auto-dispatched.
-  CONSULTATION_AGENTS = %w[plastic-advisor plastic-faux-advisor].freeze
-
-  # Codex reasoning-effort per tier alias (intent 102a). model_reasoning_effort is a
-  # depth-of-thinking dial independent of model selection (181 line 317-318), so mapping
-  # the tier here never encodes a rotting Codex model id (116 D1). opus is the deepest
-  # reasoning tier -> the deepest generally-safe effort (high, not the model-dependent
-  # xhigh); sonnet the mid execution tier -> medium; haiku the lightest -> low. minimal is
-  # unused.
+  # dispatched by the auto pipeline, not part of TIER_DEFAULTS. Codex installs
+  # generate both advisors with Astra and preserve each role's effort.
+  CONSULTATION_AGENTS = %w[plastic-primary-advisor plastic-secondary-advisor].freeze
+
+  SHIPPED_MODEL_DEFAULTS = TIER_DEFAULTS.merge(
+    "plastic-primary-advisor" => "fable",
+    "plastic-secondary-advisor" => "fable"
+  ).freeze
+
+  # Codex reasoning effort per tier alias. Model choice and reasoning effort are independent:
+  # aliases select the recommended OpenAI model, while every Plastic role starts at medium.
+  # A valid harness-scoped user override can still choose another effort for one agent.
   EFFORT_BY_ALIAS = {
-    "opus" => "high",
+    "opus" => "medium",
     "sonnet" => "medium",
-    "haiku" => "low"
+    "haiku" => "medium"
+  }.freeze
+
+  DEFAULT_EFFORT = "medium"
+
+  SHIPPED_EFFORT_DEFAULTS = {
+    "plastic-primary-advisor" => "medium",
+    "plastic-secondary-advisor" => "high"
   }.freeze
 
   # Codex model id per tier alias (intent 186). Codex has NO vendor alias layer: every model id
@@ -58,17 +60,26 @@ module AgentModels
   # centralizing every id in ONE map: Plastic owns the alias, so per-role identity costs a single
   # line to refresh on a Codex deprecation plus a Plastic release, and no per-role file carries a
   # raw id. opus (deepest reasoning tier) -> the flagship Sol; sonnet (execution tier) -> the
-  # balanced Terra; haiku (lightest) -> the fast/cheap Luna. Paired with EFFORT_BY_ALIAS so
-  # reasoning roles get a stronger model AND higher effort than executors. This is a shipped
-  # DEFAULT, fully overridable via agents.models.codex.<name>.
+  # balanced Terra; haiku (lightest) -> the fast/cheap Luna. This is a shipped
+  # DEFAULT, fully overridable via agents.models.codex.<name>. Model strength does not change
+  # effort: all three aliases use medium unless agents.efforts.codex.<name> overrides it.
   CODEX_MODEL_BY_ALIAS = {
     "opus" => "gpt-5.6-sol",
     "sonnet" => "gpt-5.6-terra",
     "haiku" => "gpt-5.6-luna"
   }.freeze
 
+  CODEX_MODEL_BY_AGENT = {
+    "plastic-primary-advisor" => "gpt-6-astra",
+    "plastic-secondary-advisor" => "gpt-6-astra"
+  }.freeze
+
   module_function
 
+  def shipped_effort_for(agent)
+    SHIPPED_EFFORT_DEFAULTS.fetch(agent.to_s, DEFAULT_EFFORT)
+  end
+
   # Pull { basename => model } out of a loaded config hash's `agents.models`
   # section, scoped to `harness` ("claude" or "codex"), tolerating a missing or
   # malformed shape. `agents.models` can mix two shapes: legacy FLAT scalar
@@ -124,4 +135,11 @@ module AgentModels
   def codex_model_for(value)
     CODEX_MODEL_BY_ALIAS[value.to_s]
   end
+
+  def shipped_model_for(agent, harness: "claude")
+    value = SHIPPED_MODEL_DEFAULTS[agent.to_s]
+    return value unless harness.to_s == "codex"
+
+    CODEX_MODEL_BY_AGENT[agent.to_s] || codex_model_for(value) || value
+  end
 end

package/scripts/lib/codex_adapter.rb

@@ -70,15 +70,16 @@ module CodexAdapter
   # worktree, never `~/.plastic`), `--output-last-message` at the path the
   # caller names, and a bare `-` so the prompt is read from stdin (row 6.9) -
   # the node input itself never rides in this array.
-  def build_argv(kind:, worktree:, output_last_message:)
-    [
+  def build_argv(kind:, worktree:, output_last_message:, model: nil, effort: nil)
+    argv = [
       "codex", "exec",
       "-C", worktree.to_s,
       "--sandbox", sandbox_mode(kind),
       *add_dir_args(kind: kind, worktree: worktree),
-      "--output-last-message", output_last_message.to_s,
-      "-",
     ]
+    argv += ["--model", model.to_s] unless model.to_s.strip.empty?
+    argv += ["--config", %(model_reasoning_effort="#{effort}")] unless effort.to_s.strip.empty?
+    argv + ["--output-last-message", output_last_message.to_s, "-"]
   end
 
   # matrix row 6.23: the wall-clock bound comes from the KIND's own lease

package/scripts/lib/graph_measure_models.rb

@@ -24,7 +24,7 @@ require_relative "agent_models"
 # through `RunnerPolicy.model_for(kind, config:)`, the one place that maps a
 # node's kind to the role the runner actually dispatched it against - never
 # a hand-rolled kind-to-role table, and never `AgentModels::TIER_DEFAULTS`
-# directly, which carries no `plastic-advisor` entry at all (that agent is a
+# directly, which carries no `plastic-primary-advisor` entry at all (that agent is a
 # consultation agent, never auto-dispatched). Adds no second parser:
 # `NodeLedger` owns the transition-line format, `NodeFile` owns the kind
 # envelope and the `KIND_PREFIX` fallback, `RunnerPolicy`/`AgentModels` own