:hedgehog: PostHog is the leading platform for building self-driving products. Our developer tools – AI observability, analytics, session replay, flags, experiments, error tracking, logs, and more – capture all the context agents need to diagnose problems, uncover opportunities, and ship fixes. Steer it all from Slack, web, desktop, or the MCP.
Moving from posthog-cli to hog
[!NOTE] This guide is a draft. hog isn't published yet, and the guide will move to posthog.com when it is.
hog is version 1 of the PostHog command line. It does what posthog-cli does, with new command names, HOG_ environment variables, and output that agents can read.
Most moves take three steps: install hog, rename the environment variables, and rewrite each command with the tables below.
What stays the same
- Your login. Run
hog auth migrateonce. It moves the keyposthog-cli loginsaved in~/.posthog/credentials.jsoninto hog's store, so you don't have to log in again. It checks the key with PostHog first, and a key PostHog rejects stays out of hog's store. hog doesn't read that file otherwise and never changes it. - Uploads. hog sends the same bytes and content hashes, so either CLI can re-upload what the other uploaded. Chunk IDs and releases don't change.
- posthog.json. hog reads the
languagesentriesposthog-cli exp schemawrites, and the endpoint YAML filesposthog-cli exp endpointswrites. - Global flags.
--host,--dotenv-file,--no-fail,--skip-ssl-verificationand--rate-limitwork as before, before or after the command.
Environment variables
hog reads only HOG_ variables. Rename these:
| posthog-cli | hog |
|---|---|
POSTHOG_CLI_API_KEY, POSTHOG_CLI_TOKEN |
HOG_PERSONAL_API_KEY |
POSTHOG_CLI_PROJECT_ID, POSTHOG_CLI_ENV_ID |
HOG_PROJECT_ID |
POSTHOG_CLI_HOST |
HOG_HOST |
POSTHOG_CLI_DOTENV_FILE |
HOG_DOTENV_FILE |
POSTHOG_CLI_DRY_RUN |
HOG_DRY_RUN |
POSTHOG_CLIENT_RATE_LIMIT |
HOG_RATE_LIMIT |
POSTHOG_CLI_SOURCEMAP_UPLOAD_CONCURRENCY |
HOG_UPLOAD_CONCURRENCY |
POSTHOG_RELEASE_MODE |
HOG_RELEASE_MODE |
POSTHOG_HOME |
HOG_HOME |
A dotenv file that --dotenv-file names needs the new names too.
In CI, you can set both sets of names while some jobs still run posthog-cli.
Commands
| posthog-cli | hog |
|---|---|
login |
auth login, plus auth status and auth logout |
sourcemap inject --directory <dir> |
error-tracking inject <dir> |
sourcemap upload --directory <dir> |
error-tracking upload <dir> |
sourcemap process --directory <dir> |
error-tracking process <dir> |
hermes inject --directory <dir> |
error-tracking inject <dir> |
hermes clone -m <file1> -c <file2> |
error-tracking clone <file1> <file2> |
hermes upload --directory <dir> |
error-tracking upload <dir> |
dsym upload --directory <dir> |
error-tracking upload <dir> |
proguard upload --path <file> --map-id <id> |
error-tracking upload <file> --map-id <id> |
symbol-sets upload --directory <dir> |
error-tracking upload <dir> |
symbol-sets download --id <id> -o <dir> |
error-tracking download <id> --output-dir <dir> |
symbol-sets download --ref <ref> |
error-tracking download <ref>, with --ref when the ref looks like an ID |
symbol-sets extract <file> -o <dir> |
error-tracking extract <file> --output-dir <dir> |
release resolve |
error-tracking release -o text |
release resolve --json |
error-tracking release -o json |
exp query run <query> |
query run <query> |
exp query check <query> |
query check <query> |
exp endpoints list, get, open, run, diff |
endpoints list, get, open, run, diff. list shows one page, and --offset picks the next |
exp endpoints pull -o <dir> |
endpoints pull --to <dir> |
exp endpoints push |
endpoints push |
exp schema pull -o <file> |
schema pull --file <file> |
exp schema status |
schema status |
api tools, search, info, schema, call |
api tools, search, info, schema, call, with -o json for --json |
hog et is short for hog error-tracking.
hog error-tracking upload detects what each path holds, so one command uploads JavaScript, Hermes, ProGuard, dSYM and ELF symbols, and --kind skips detection.
Run hog <command> --help for every flag, or hog commands -o json for all of them at once.
Not in hog yet, so keep posthog-cli for them:
update. It becomeshog upgradewhen hog ships.api skillandapi agents-md. They come back as top-level commands.exp logs import,exp task, andexp query editor.
hog lists these and prints the posthog-cli command to run instead.
Flags that changed
- Paths are arguments.
--directory,--path,-mand-cbecome positional arguments, and one command takes several paths. -ois the output format. It takestextorjson. posthog-cli's-opaths become--toinendpoints pull,--fileinschema pull, and--output-dirindownloadandextract.--include,--excludeand--stdinare gone. List the paths as arguments, or keep them inerrorTracking.sourcesinposthog.json, whereignoreglobs skip files.--skip-release-on-failis gone. hog always retries without the release when PostHog says a chunk belongs to another release, as posthog-cli did.--main-dsymis gone. hog takes the release from--release-name,--release-version,--info-plist,posthog.jsonor git.--debugis global. It prints each request to stderr for every command.
Dry runs
posthog-cli's --dry-run skipped uploads and release without doing anything.
hog's --dry-run does the work and shows the result without changing anything: uploads list each symbol set with its content hash, and schema pull, endpoints pull and push show what they would change.
HOG_DRY_RUN=1 turns on --dry-run for every command that has one. Commands that only read run as usual.
auth login and auth logout have no dry run, so they fail when HOG_DRY_RUN is set.
Output
hog prints tables and colors for a person at a terminal, and JSON when an agent runs it, when stdout is a pipe, or with -o json.
A script that reads hog's output should pass -o text or -o json, so the format doesn't depend on where it runs:
# before
RELEASE_ID=$(posthog-cli release resolve)
# after
RELEASE_ID=$(hog error-tracking release -o text)
Errors go to stderr, as JSON with a stable code in JSON mode.
Exit codes are 0 on success, 1 when a command fails or reports a problem, and 2 for usage errors.
--no-fail turns 1 into 0, and leaves 2 alone.
posthog.json
hog reads more from posthog.json than posthog-cli did: the host, error tracking sources and release, the endpoints folder, and defaults for --no-fail and --rate-limit.
When the file is invalid, every command stops except hog config check, which lists the problems.
posthog-cli rewrites posthog.json with only languages when it pulls a schema, so it drops hog's settings.
Run only hog's schema pull once a project uses hog settings.