Module output 

Human output: the dry-run and --verbose report (spec §26, SP1 design §7.4, SP2 design §7.3), the repository command reports (SP3 design §7.5) and redaction (spec §22). JSON output (spec §32) is built in json (SP5b).

Re-exports

pub use crate::exe::ExecutableStatus;

Enums

ReportMode

Which report is rendered.

Shown 🔒

How one opaque argument is shown (spec §26 “Sensitive values must be redacted”, §36).

Constants

LABEL_WIDTH 🔒

SENSITIVE_NAME_PARTS 🔒

Substrings that mark an override variable as secret-bearing, whatever the adapter declared.

Functions

agent_mappings_remain 🔒

The note that agent mappings survive a repository-level unlink, or nothing when there are none.

agents_lines

agents (SP5a design §3): a padded block per entry — the agent, the executable status, the mechanism, the support level, then the evidence entry’s measured upstream version with its verification date (a probe of whatever happens to be installed would say nothing about the matrix it pairs with), and the full capability matrix without which a support level is never shown (SP4 design §6.1). An empty line separates the entries; there is none after the last.

capability_label

How a capability is spelled. One vocabulary covers the report and the hedge (SP4 design §7.2).

capability_lines

The capability-matrix block of the report: every capability in Capability::ALL, in that fixed order, each followed by its basis. The matrix is unconditional: SP4 design §6.1 rests on a support level never being shown without it, so every block that shows a support level (report_lines, agents_lines) renders this whole, without a leading or a trailing line of its own.

classify 🔒

create_line

create (SP5a design §3).

delete_line

delete (SP5a design §3): the stdout line of each outcome. A profile that vanished before the locked delete ran reports the same does-not-exist line as an advisory miss, because that is the state the user asked for.

doctor_lines

The doctor report (SP5c design §7): records grouped under host, repositories and one agent <id> heading per agent, then the summary line.

has_sensitive_part 🔒

is_dotted_identifier 🔒

A_1, mcp_servers.gh.env.GITHUB_TOKEN or mcp_servers.chrome-devtools.http_headers.X-Api-Key: TOML bare keys (letters, digits, _, -) joined by dots, as Codex -c dotted paths are.

is_sensitive 🔒

labeled 🔒

label: padded to the report column, then value.

link_line

link (SP3 design §7.5).

list_lines

list (SP5a design §3): only the names of the rows whose agents contain agent, one per line. The result may be empty; the CLI layer decides what that means for stderr.

missing_profile_line

The stdout line for a deleting command whose profile is not there.

note

A note: line.

or_none 🔒

origin_word 🔒

Where a launch’s executable came from, spelled as the report does.

profiles_lines

profiles (SP5a design §3): one line per row, the profile name padded to the longest name so the agent columns align; an empty listing is the single line no profiles. A profile with no agent materialized is shown as (none materialized) rather than an empty column.

render_arg 🔒

render_args 🔒

Renders the opaque arguments for the report. Only the report is redacted; the launched arguments never change. Redaction is a shallow name rule, never a parser: a --option whose name holds a sensitive part hides its value (--api-key=<redacted>, or the next argument); a NAME=value whose (possibly dotted) NAME holds one hides the value (mcp_servers.gh.env.GITHUB_TOKEN=<redacted>); a Name: value header whose name holds one hides the value (Authorization: <redacted>). Every argument is scanned, including after a --, and boolean-looking names get no exemption, because hiding too much only costs readability.

report_lines

The report lines, without trailing newlines.

repositories_lines

repositories (SP5b design §4): a block per stored mapping, a blank line between blocks. The key is printed as stored.

resolve_lines

<agent> resolve (SP3 design §7.5).

sensitive_value_prefix 🔒

The shown prefix when text is a NAME=value assignment or a Name: value header whose name holds a sensitive part: NAME= or Name: .

shell_word 🔒

A path to paste into a command: in double quotes when it contains whitespace, which POSIX shells, PowerShell and cmd all read as one argument.

state_label

How a capability state is spelled: lower case, with spaces (SP4 design §7.1).

status_lines

status and <agent> status: agents holds one resolution per agent line and, for <agent> status, its presence: value (SP3 design §7.5).

support_hedge

The one-line stderr hedge for an adapter that is not Proven, or None when it is (SP4 design §7.2).

support_label

How a support level is spelled in the report and in the launch hedge (SP4 design §7.1).

unlink_lines

unlink (SP3 design §7.5). config is the configuration read before the command, keys the candidates.